diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,12864 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9975485904395027,
+  "eval_steps": 500,
+  "global_step": 2139,
+  "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": 1.4057,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2797901240490773e-05,
+      "loss": 1.2147,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0284193554110643e-05,
+      "loss": 1.2292,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5595802480981545e-05,
+      "loss": 1.0036,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9715806445889356e-05,
+      "loss": 0.975,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.308209479460142e-05,
+      "loss": 0.8997,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.592825103949889e-05,
+      "loss": 0.8972,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.839370372147232e-05,
+      "loss": 0.8793,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.056838710822129e-05,
+      "loss": 0.8286,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.251370768638013e-05,
+      "loss": 0.8187,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.427346420355127e-05,
+      "loss": 0.7637,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.587999603509219e-05,
+      "loss": 0.8377,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.735786205915921e-05,
+      "loss": 0.7566,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8726152279989664e-05,
+      "loss": 0.747,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9999999999999996e-05,
+      "loss": 0.7104,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-05,
+      "loss": 0.6518,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997645951035782e-05,
+      "loss": 0.6564,
+      "step": 17
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9952919020715633e-05,
+      "loss": 0.6582,
+      "step": 18
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.992937853107345e-05,
+      "loss": 0.652,
+      "step": 19
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9905838041431265e-05,
+      "loss": 0.6791,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9882297551789083e-05,
+      "loss": 0.6092,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9858757062146896e-05,
+      "loss": 0.6358,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9835216572504714e-05,
+      "loss": 0.6248,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.981167608286253e-05,
+      "loss": 0.6092,
+      "step": 24
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.978813559322034e-05,
+      "loss": 0.593,
+      "step": 25
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.976459510357816e-05,
+      "loss": 0.6363,
+      "step": 26
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.974105461393597e-05,
+      "loss": 0.626,
+      "step": 27
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.971751412429379e-05,
+      "loss": 0.6142,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.96939736346516e-05,
+      "loss": 0.6425,
+      "step": 29
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.967043314500942e-05,
+      "loss": 0.6317,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.964689265536723e-05,
+      "loss": 0.6096,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.962335216572505e-05,
+      "loss": 0.6382,
+      "step": 32
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.959981167608286e-05,
+      "loss": 0.6097,
+      "step": 33
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.957627118644068e-05,
+      "loss": 0.5644,
+      "step": 34
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9552730696798494e-05,
+      "loss": 0.6072,
+      "step": 35
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.952919020715631e-05,
+      "loss": 0.5821,
+      "step": 36
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9505649717514125e-05,
+      "loss": 0.587,
+      "step": 37
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9482109227871944e-05,
+      "loss": 0.6077,
+      "step": 38
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.945856873822976e-05,
+      "loss": 0.6201,
+      "step": 39
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9435028248587575e-05,
+      "loss": 0.6543,
+      "step": 40
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9411487758945394e-05,
+      "loss": 0.6235,
+      "step": 41
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9387947269303206e-05,
+      "loss": 0.5676,
+      "step": 42
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.936440677966102e-05,
+      "loss": 0.6004,
+      "step": 43
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.934086629001883e-05,
+      "loss": 0.6102,
+      "step": 44
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.931732580037665e-05,
+      "loss": 0.5754,
+      "step": 45
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.929378531073446e-05,
+      "loss": 0.6097,
+      "step": 46
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.927024482109228e-05,
+      "loss": 0.6076,
+      "step": 47
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.924670433145009e-05,
+      "loss": 0.6022,
+      "step": 48
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.922316384180791e-05,
+      "loss": 0.5839,
+      "step": 49
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.919962335216573e-05,
+      "loss": 0.5503,
+      "step": 50
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.917608286252354e-05,
+      "loss": 0.5704,
+      "step": 51
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.915254237288136e-05,
+      "loss": 0.6262,
+      "step": 52
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9129001883239174e-05,
+      "loss": 0.5648,
+      "step": 53
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.910546139359699e-05,
+      "loss": 0.5426,
+      "step": 54
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9081920903954805e-05,
+      "loss": 0.6006,
+      "step": 55
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9058380414312623e-05,
+      "loss": 0.5479,
+      "step": 56
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9034839924670436e-05,
+      "loss": 0.5687,
+      "step": 57
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9011299435028255e-05,
+      "loss": 0.5625,
+      "step": 58
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.898775894538607e-05,
+      "loss": 0.5887,
+      "step": 59
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.896421845574388e-05,
+      "loss": 0.5577,
+      "step": 60
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.89406779661017e-05,
+      "loss": 0.5741,
+      "step": 61
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.891713747645951e-05,
+      "loss": 0.5621,
+      "step": 62
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.889359698681733e-05,
+      "loss": 0.5754,
+      "step": 63
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.887005649717514e-05,
+      "loss": 0.5832,
+      "step": 64
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.884651600753296e-05,
+      "loss": 0.569,
+      "step": 65
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.882297551789077e-05,
+      "loss": 0.5904,
+      "step": 66
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.879943502824859e-05,
+      "loss": 0.5779,
+      "step": 67
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.87758945386064e-05,
+      "loss": 0.5798,
+      "step": 68
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.875235404896422e-05,
+      "loss": 0.5908,
+      "step": 69
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8728813559322034e-05,
+      "loss": 0.5865,
+      "step": 70
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.870527306967985e-05,
+      "loss": 0.5794,
+      "step": 71
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8681732580037665e-05,
+      "loss": 0.5683,
+      "step": 72
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8658192090395484e-05,
+      "loss": 0.5593,
+      "step": 73
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.86346516007533e-05,
+      "loss": 0.5761,
+      "step": 74
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8611111111111115e-05,
+      "loss": 0.5548,
+      "step": 75
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8587570621468934e-05,
+      "loss": 0.5946,
+      "step": 76
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8564030131826746e-05,
+      "loss": 0.5679,
+      "step": 77
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.854048964218456e-05,
+      "loss": 0.5444,
+      "step": 78
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.851694915254237e-05,
+      "loss": 0.5754,
+      "step": 79
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.849340866290019e-05,
+      "loss": 0.6216,
+      "step": 80
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8469868173258e-05,
+      "loss": 0.544,
+      "step": 81
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.844632768361582e-05,
+      "loss": 0.5316,
+      "step": 82
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.842278719397363e-05,
+      "loss": 0.5611,
+      "step": 83
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.839924670433145e-05,
+      "loss": 0.5588,
+      "step": 84
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.837570621468927e-05,
+      "loss": 0.5425,
+      "step": 85
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.835216572504708e-05,
+      "loss": 0.5337,
+      "step": 86
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.83286252354049e-05,
+      "loss": 0.5697,
+      "step": 87
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8305084745762714e-05,
+      "loss": 0.5619,
+      "step": 88
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.828154425612053e-05,
+      "loss": 0.5504,
+      "step": 89
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8258003766478345e-05,
+      "loss": 0.5475,
+      "step": 90
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8234463276836164e-05,
+      "loss": 0.5833,
+      "step": 91
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8210922787193976e-05,
+      "loss": 0.5053,
+      "step": 92
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8187382297551795e-05,
+      "loss": 0.574,
+      "step": 93
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.816384180790961e-05,
+      "loss": 0.5445,
+      "step": 94
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.814030131826742e-05,
+      "loss": 0.5832,
+      "step": 95
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.811676082862524e-05,
+      "loss": 0.5683,
+      "step": 96
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.809322033898305e-05,
+      "loss": 0.5552,
+      "step": 97
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.806967984934087e-05,
+      "loss": 0.5495,
+      "step": 98
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.804613935969868e-05,
+      "loss": 0.5712,
+      "step": 99
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.80225988700565e-05,
+      "loss": 0.5343,
+      "step": 100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.799905838041431e-05,
+      "loss": 0.5633,
+      "step": 101
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.797551789077213e-05,
+      "loss": 0.5335,
+      "step": 102
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.795197740112994e-05,
+      "loss": 0.512,
+      "step": 103
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.792843691148776e-05,
+      "loss": 0.5467,
+      "step": 104
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7904896421845574e-05,
+      "loss": 0.5401,
+      "step": 105
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.788135593220339e-05,
+      "loss": 0.5246,
+      "step": 106
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.785781544256121e-05,
+      "loss": 0.5588,
+      "step": 107
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7834274952919024e-05,
+      "loss": 0.5512,
+      "step": 108
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.781073446327684e-05,
+      "loss": 0.5786,
+      "step": 109
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7787193973634655e-05,
+      "loss": 0.5858,
+      "step": 110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7763653483992474e-05,
+      "loss": 0.577,
+      "step": 111
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7740112994350286e-05,
+      "loss": 0.5599,
+      "step": 112
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.77165725047081e-05,
+      "loss": 0.5502,
+      "step": 113
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.769303201506591e-05,
+      "loss": 0.5437,
+      "step": 114
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.766949152542373e-05,
+      "loss": 0.5473,
+      "step": 115
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.764595103578154e-05,
+      "loss": 0.5675,
+      "step": 116
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.762241054613936e-05,
+      "loss": 0.5731,
+      "step": 117
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.759887005649718e-05,
+      "loss": 0.5558,
+      "step": 118
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.757532956685499e-05,
+      "loss": 0.5408,
+      "step": 119
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.755178907721281e-05,
+      "loss": 0.5515,
+      "step": 120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.752824858757062e-05,
+      "loss": 0.556,
+      "step": 121
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.750470809792844e-05,
+      "loss": 0.5358,
+      "step": 122
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7481167608286254e-05,
+      "loss": 0.5712,
+      "step": 123
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.745762711864407e-05,
+      "loss": 0.5324,
+      "step": 124
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7434086629001885e-05,
+      "loss": 0.5506,
+      "step": 125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7410546139359704e-05,
+      "loss": 0.568,
+      "step": 126
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7387005649717516e-05,
+      "loss": 0.5182,
+      "step": 127
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7363465160075335e-05,
+      "loss": 0.525,
+      "step": 128
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.733992467043315e-05,
+      "loss": 0.5194,
+      "step": 129
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7316384180790966e-05,
+      "loss": 0.5299,
+      "step": 130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.729284369114878e-05,
+      "loss": 0.547,
+      "step": 131
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.726930320150659e-05,
+      "loss": 0.5457,
+      "step": 132
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.724576271186441e-05,
+      "loss": 0.5269,
+      "step": 133
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.722222222222222e-05,
+      "loss": 0.5524,
+      "step": 134
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.719868173258004e-05,
+      "loss": 0.5384,
+      "step": 135
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.717514124293785e-05,
+      "loss": 0.5277,
+      "step": 136
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.715160075329567e-05,
+      "loss": 0.5332,
+      "step": 137
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.712806026365348e-05,
+      "loss": 0.5699,
+      "step": 138
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.71045197740113e-05,
+      "loss": 0.5664,
+      "step": 139
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.7080979284369114e-05,
+      "loss": 0.5361,
+      "step": 140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.705743879472693e-05,
+      "loss": 0.5371,
+      "step": 141
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.703389830508475e-05,
+      "loss": 0.547,
+      "step": 142
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.7010357815442564e-05,
+      "loss": 0.5569,
+      "step": 143
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.698681732580038e-05,
+      "loss": 0.565,
+      "step": 144
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6963276836158195e-05,
+      "loss": 0.5493,
+      "step": 145
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6939736346516014e-05,
+      "loss": 0.5345,
+      "step": 146
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6916195856873826e-05,
+      "loss": 0.5477,
+      "step": 147
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.689265536723164e-05,
+      "loss": 0.5141,
+      "step": 148
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.686911487758945e-05,
+      "loss": 0.5262,
+      "step": 149
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.684557438794727e-05,
+      "loss": 0.567,
+      "step": 150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.682203389830508e-05,
+      "loss": 0.5475,
+      "step": 151
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.67984934086629e-05,
+      "loss": 0.5564,
+      "step": 152
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.677495291902072e-05,
+      "loss": 0.5421,
+      "step": 153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.675141242937853e-05,
+      "loss": 0.5321,
+      "step": 154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.672787193973635e-05,
+      "loss": 0.5342,
+      "step": 155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.670433145009416e-05,
+      "loss": 0.5397,
+      "step": 156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.668079096045198e-05,
+      "loss": 0.5346,
+      "step": 157
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6657250470809794e-05,
+      "loss": 0.5784,
+      "step": 158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.663370998116761e-05,
+      "loss": 0.5604,
+      "step": 159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6610169491525425e-05,
+      "loss": 0.5319,
+      "step": 160
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6586629001883244e-05,
+      "loss": 0.5441,
+      "step": 161
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6563088512241056e-05,
+      "loss": 0.5461,
+      "step": 162
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6539548022598875e-05,
+      "loss": 0.5602,
+      "step": 163
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.651600753295669e-05,
+      "loss": 0.5512,
+      "step": 164
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6492467043314506e-05,
+      "loss": 0.5467,
+      "step": 165
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.646892655367232e-05,
+      "loss": 0.5319,
+      "step": 166
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.644538606403013e-05,
+      "loss": 0.5536,
+      "step": 167
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.642184557438795e-05,
+      "loss": 0.5287,
+      "step": 168
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.639830508474576e-05,
+      "loss": 0.532,
+      "step": 169
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.637476459510358e-05,
+      "loss": 0.5282,
+      "step": 170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.635122410546139e-05,
+      "loss": 0.543,
+      "step": 171
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.632768361581921e-05,
+      "loss": 0.5057,
+      "step": 172
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.630414312617702e-05,
+      "loss": 0.5504,
+      "step": 173
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.628060263653484e-05,
+      "loss": 0.5599,
+      "step": 174
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.625706214689266e-05,
+      "loss": 0.5282,
+      "step": 175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.623352165725047e-05,
+      "loss": 0.5481,
+      "step": 176
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.620998116760829e-05,
+      "loss": 0.5066,
+      "step": 177
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6186440677966104e-05,
+      "loss": 0.5289,
+      "step": 178
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.616290018832392e-05,
+      "loss": 0.5541,
+      "step": 179
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6139359698681735e-05,
+      "loss": 0.5009,
+      "step": 180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6115819209039554e-05,
+      "loss": 0.5381,
+      "step": 181
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6092278719397366e-05,
+      "loss": 0.5443,
+      "step": 182
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.606873822975518e-05,
+      "loss": 0.5411,
+      "step": 183
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.6045197740113e-05,
+      "loss": 0.5237,
+      "step": 184
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.602165725047081e-05,
+      "loss": 0.5149,
+      "step": 185
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.599811676082863e-05,
+      "loss": 0.5043,
+      "step": 186
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.597457627118644e-05,
+      "loss": 0.485,
+      "step": 187
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.595103578154426e-05,
+      "loss": 0.5076,
+      "step": 188
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.592749529190207e-05,
+      "loss": 0.5283,
+      "step": 189
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.590395480225989e-05,
+      "loss": 0.5453,
+      "step": 190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.58804143126177e-05,
+      "loss": 0.5294,
+      "step": 191
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.585687382297552e-05,
+      "loss": 0.5552,
+      "step": 192
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5833333333333334e-05,
+      "loss": 0.5221,
+      "step": 193
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.580979284369115e-05,
+      "loss": 0.5405,
+      "step": 194
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5786252354048965e-05,
+      "loss": 0.5433,
+      "step": 195
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5762711864406784e-05,
+      "loss": 0.5409,
+      "step": 196
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.57391713747646e-05,
+      "loss": 0.539,
+      "step": 197
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5715630885122415e-05,
+      "loss": 0.5429,
+      "step": 198
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5692090395480234e-05,
+      "loss": 0.571,
+      "step": 199
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5668549905838046e-05,
+      "loss": 0.5324,
+      "step": 200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.564500941619586e-05,
+      "loss": 0.5084,
+      "step": 201
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.562146892655367e-05,
+      "loss": 0.5219,
+      "step": 202
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.559792843691149e-05,
+      "loss": 0.5351,
+      "step": 203
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.55743879472693e-05,
+      "loss": 0.5232,
+      "step": 204
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.555084745762712e-05,
+      "loss": 0.5233,
+      "step": 205
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.552730696798493e-05,
+      "loss": 0.5108,
+      "step": 206
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.550376647834275e-05,
+      "loss": 0.5359,
+      "step": 207
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.548022598870056e-05,
+      "loss": 0.4744,
+      "step": 208
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.545668549905838e-05,
+      "loss": 0.5011,
+      "step": 209
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.54331450094162e-05,
+      "loss": 0.5004,
+      "step": 210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.540960451977401e-05,
+      "loss": 0.5366,
+      "step": 211
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.538606403013183e-05,
+      "loss": 0.532,
+      "step": 212
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5362523540489644e-05,
+      "loss": 0.5343,
+      "step": 213
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.533898305084746e-05,
+      "loss": 0.5279,
+      "step": 214
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5315442561205275e-05,
+      "loss": 0.5345,
+      "step": 215
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5291902071563094e-05,
+      "loss": 0.5374,
+      "step": 216
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5268361581920906e-05,
+      "loss": 0.5152,
+      "step": 217
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.524482109227872e-05,
+      "loss": 0.5429,
+      "step": 218
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.522128060263654e-05,
+      "loss": 0.5623,
+      "step": 219
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.519774011299435e-05,
+      "loss": 0.535,
+      "step": 220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.517419962335217e-05,
+      "loss": 0.5222,
+      "step": 221
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.515065913370998e-05,
+      "loss": 0.5231,
+      "step": 222
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.51271186440678e-05,
+      "loss": 0.5089,
+      "step": 223
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.510357815442561e-05,
+      "loss": 0.5018,
+      "step": 224
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.508003766478343e-05,
+      "loss": 0.5355,
+      "step": 225
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.505649717514124e-05,
+      "loss": 0.5074,
+      "step": 226
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.503295668549906e-05,
+      "loss": 0.5243,
+      "step": 227
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.5009416195856874e-05,
+      "loss": 0.5242,
+      "step": 228
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.498587570621469e-05,
+      "loss": 0.5032,
+      "step": 229
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4962335216572505e-05,
+      "loss": 0.5233,
+      "step": 230
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4938794726930324e-05,
+      "loss": 0.5363,
+      "step": 231
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.491525423728814e-05,
+      "loss": 0.5548,
+      "step": 232
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4891713747645955e-05,
+      "loss": 0.5069,
+      "step": 233
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4868173258003774e-05,
+      "loss": 0.5395,
+      "step": 234
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4844632768361586e-05,
+      "loss": 0.5078,
+      "step": 235
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.48210922787194e-05,
+      "loss": 0.5215,
+      "step": 236
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.479755178907721e-05,
+      "loss": 0.5312,
+      "step": 237
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.477401129943503e-05,
+      "loss": 0.4965,
+      "step": 238
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.475047080979284e-05,
+      "loss": 0.5196,
+      "step": 239
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.472693032015066e-05,
+      "loss": 0.5517,
+      "step": 240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.470338983050847e-05,
+      "loss": 0.5315,
+      "step": 241
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.467984934086629e-05,
+      "loss": 0.5175,
+      "step": 242
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.465630885122411e-05,
+      "loss": 0.5291,
+      "step": 243
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.463276836158192e-05,
+      "loss": 0.5356,
+      "step": 244
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.460922787193974e-05,
+      "loss": 0.5246,
+      "step": 245
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4585687382297553e-05,
+      "loss": 0.5433,
+      "step": 246
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.456214689265537e-05,
+      "loss": 0.5304,
+      "step": 247
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4538606403013184e-05,
+      "loss": 0.5361,
+      "step": 248
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4515065913371e-05,
+      "loss": 0.5168,
+      "step": 249
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4491525423728816e-05,
+      "loss": 0.5213,
+      "step": 250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4467984934086634e-05,
+      "loss": 0.5284,
+      "step": 251
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4444444444444447e-05,
+      "loss": 0.4989,
+      "step": 252
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.442090395480226e-05,
+      "loss": 0.5273,
+      "step": 253
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.439736346516008e-05,
+      "loss": 0.4986,
+      "step": 254
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.437382297551789e-05,
+      "loss": 0.5167,
+      "step": 255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.435028248587571e-05,
+      "loss": 0.5102,
+      "step": 256
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.432674199623352e-05,
+      "loss": 0.5468,
+      "step": 257
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.430320150659134e-05,
+      "loss": 0.5178,
+      "step": 258
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.427966101694915e-05,
+      "loss": 0.5234,
+      "step": 259
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.425612052730697e-05,
+      "loss": 0.5263,
+      "step": 260
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.423258003766478e-05,
+      "loss": 0.5282,
+      "step": 261
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.42090395480226e-05,
+      "loss": 0.5252,
+      "step": 262
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.4185499058380414e-05,
+      "loss": 0.5241,
+      "step": 263
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.416195856873823e-05,
+      "loss": 0.4965,
+      "step": 264
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.413841807909605e-05,
+      "loss": 0.5343,
+      "step": 265
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.4114877589453864e-05,
+      "loss": 0.5301,
+      "step": 266
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.409133709981168e-05,
+      "loss": 0.5093,
+      "step": 267
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.4067796610169495e-05,
+      "loss": 0.5346,
+      "step": 268
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.4044256120527314e-05,
+      "loss": 0.5146,
+      "step": 269
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.4020715630885126e-05,
+      "loss": 0.5332,
+      "step": 270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.399717514124294e-05,
+      "loss": 0.4683,
+      "step": 271
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.397363465160075e-05,
+      "loss": 0.5175,
+      "step": 272
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.395009416195857e-05,
+      "loss": 0.5442,
+      "step": 273
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.392655367231638e-05,
+      "loss": 0.5207,
+      "step": 274
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.39030131826742e-05,
+      "loss": 0.5314,
+      "step": 275
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.387947269303202e-05,
+      "loss": 0.5388,
+      "step": 276
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.385593220338983e-05,
+      "loss": 0.5195,
+      "step": 277
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.383239171374765e-05,
+      "loss": 0.4813,
+      "step": 278
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.380885122410546e-05,
+      "loss": 0.5516,
+      "step": 279
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.378531073446328e-05,
+      "loss": 0.5073,
+      "step": 280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3761770244821093e-05,
+      "loss": 0.5279,
+      "step": 281
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.373822975517891e-05,
+      "loss": 0.5388,
+      "step": 282
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3714689265536725e-05,
+      "loss": 0.5254,
+      "step": 283
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3691148775894543e-05,
+      "loss": 0.5387,
+      "step": 284
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3667608286252356e-05,
+      "loss": 0.5194,
+      "step": 285
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3644067796610175e-05,
+      "loss": 0.529,
+      "step": 286
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.362052730696799e-05,
+      "loss": 0.5072,
+      "step": 287
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3596986817325806e-05,
+      "loss": 0.512,
+      "step": 288
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.357344632768362e-05,
+      "loss": 0.5106,
+      "step": 289
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.354990583804143e-05,
+      "loss": 0.5229,
+      "step": 290
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.352636534839925e-05,
+      "loss": 0.5267,
+      "step": 291
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.350282485875706e-05,
+      "loss": 0.5067,
+      "step": 292
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.347928436911488e-05,
+      "loss": 0.5365,
+      "step": 293
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.345574387947269e-05,
+      "loss": 0.5212,
+      "step": 294
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.343220338983051e-05,
+      "loss": 0.5145,
+      "step": 295
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.340866290018832e-05,
+      "loss": 0.5129,
+      "step": 296
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.338512241054614e-05,
+      "loss": 0.5312,
+      "step": 297
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3361581920903954e-05,
+      "loss": 0.4889,
+      "step": 298
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.333804143126177e-05,
+      "loss": 0.5098,
+      "step": 299
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.331450094161959e-05,
+      "loss": 0.5025,
+      "step": 300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3290960451977404e-05,
+      "loss": 0.4782,
+      "step": 301
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.326741996233522e-05,
+      "loss": 0.5126,
+      "step": 302
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3243879472693035e-05,
+      "loss": 0.4867,
+      "step": 303
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.3220338983050854e-05,
+      "loss": 0.5031,
+      "step": 304
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.3196798493408666e-05,
+      "loss": 0.4891,
+      "step": 305
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.317325800376648e-05,
+      "loss": 0.5257,
+      "step": 306
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.314971751412429e-05,
+      "loss": 0.4937,
+      "step": 307
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.312617702448211e-05,
+      "loss": 0.5191,
+      "step": 308
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.310263653483992e-05,
+      "loss": 0.5027,
+      "step": 309
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.307909604519774e-05,
+      "loss": 0.5229,
+      "step": 310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.305555555555556e-05,
+      "loss": 0.5071,
+      "step": 311
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.303201506591337e-05,
+      "loss": 0.5012,
+      "step": 312
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.300847457627119e-05,
+      "loss": 0.4666,
+      "step": 313
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2984934086629e-05,
+      "loss": 0.5006,
+      "step": 314
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.296139359698682e-05,
+      "loss": 0.5524,
+      "step": 315
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2937853107344634e-05,
+      "loss": 0.5026,
+      "step": 316
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.291431261770245e-05,
+      "loss": 0.5078,
+      "step": 317
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2890772128060265e-05,
+      "loss": 0.5006,
+      "step": 318
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2867231638418084e-05,
+      "loss": 0.4974,
+      "step": 319
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2843691148775896e-05,
+      "loss": 0.5088,
+      "step": 320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2820150659133715e-05,
+      "loss": 0.4814,
+      "step": 321
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.279661016949153e-05,
+      "loss": 0.492,
+      "step": 322
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2773069679849346e-05,
+      "loss": 0.5024,
+      "step": 323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.274952919020716e-05,
+      "loss": 0.5394,
+      "step": 324
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.272598870056497e-05,
+      "loss": 0.5019,
+      "step": 325
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.270244821092279e-05,
+      "loss": 0.5234,
+      "step": 326
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.26789077212806e-05,
+      "loss": 0.5241,
+      "step": 327
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.265536723163842e-05,
+      "loss": 0.4941,
+      "step": 328
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.263182674199623e-05,
+      "loss": 0.4895,
+      "step": 329
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.260828625235405e-05,
+      "loss": 0.518,
+      "step": 330
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.258474576271186e-05,
+      "loss": 0.4972,
+      "step": 331
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.256120527306968e-05,
+      "loss": 0.4895,
+      "step": 332
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.25376647834275e-05,
+      "loss": 0.4917,
+      "step": 333
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.251412429378531e-05,
+      "loss": 0.4817,
+      "step": 334
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.249058380414313e-05,
+      "loss": 0.4926,
+      "step": 335
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2467043314500944e-05,
+      "loss": 0.5047,
+      "step": 336
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.244350282485876e-05,
+      "loss": 0.4735,
+      "step": 337
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2419962335216575e-05,
+      "loss": 0.4913,
+      "step": 338
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2396421845574394e-05,
+      "loss": 0.5181,
+      "step": 339
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2372881355932206e-05,
+      "loss": 0.5149,
+      "step": 340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.234934086629002e-05,
+      "loss": 0.5222,
+      "step": 341
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.232580037664784e-05,
+      "loss": 0.513,
+      "step": 342
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.230225988700565e-05,
+      "loss": 0.5169,
+      "step": 343
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.227871939736347e-05,
+      "loss": 0.5449,
+      "step": 344
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.225517890772128e-05,
+      "loss": 0.4943,
+      "step": 345
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.22316384180791e-05,
+      "loss": 0.4925,
+      "step": 346
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.220809792843691e-05,
+      "loss": 0.4973,
+      "step": 347
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.218455743879473e-05,
+      "loss": 0.5278,
+      "step": 348
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.216101694915254e-05,
+      "loss": 0.5087,
+      "step": 349
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.213747645951036e-05,
+      "loss": 0.5051,
+      "step": 350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.2113935969868174e-05,
+      "loss": 0.5118,
+      "step": 351
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.209039548022599e-05,
+      "loss": 0.5033,
+      "step": 352
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.2066854990583805e-05,
+      "loss": 0.5261,
+      "step": 353
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.2043314500941624e-05,
+      "loss": 0.491,
+      "step": 354
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.201977401129944e-05,
+      "loss": 0.5353,
+      "step": 355
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1996233521657255e-05,
+      "loss": 0.497,
+      "step": 356
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1972693032015074e-05,
+      "loss": 0.5073,
+      "step": 357
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1949152542372886e-05,
+      "loss": 0.5118,
+      "step": 358
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.19256120527307e-05,
+      "loss": 0.4991,
+      "step": 359
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.190207156308851e-05,
+      "loss": 0.5161,
+      "step": 360
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.187853107344633e-05,
+      "loss": 0.4742,
+      "step": 361
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.185499058380414e-05,
+      "loss": 0.4865,
+      "step": 362
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.183145009416196e-05,
+      "loss": 0.4966,
+      "step": 363
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.180790960451977e-05,
+      "loss": 0.5105,
+      "step": 364
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.178436911487759e-05,
+      "loss": 0.4768,
+      "step": 365
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.17608286252354e-05,
+      "loss": 0.5323,
+      "step": 366
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.173728813559322e-05,
+      "loss": 0.5171,
+      "step": 367
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.171374764595104e-05,
+      "loss": 0.5171,
+      "step": 368
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.169020715630885e-05,
+      "loss": 0.5087,
+      "step": 369
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.4936,
+      "step": 370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1643126177024484e-05,
+      "loss": 0.5186,
+      "step": 371
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.16195856873823e-05,
+      "loss": 0.4756,
+      "step": 372
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1596045197740115e-05,
+      "loss": 0.483,
+      "step": 373
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1572504708097934e-05,
+      "loss": 0.5025,
+      "step": 374
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1548964218455746e-05,
+      "loss": 0.4834,
+      "step": 375
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.152542372881356e-05,
+      "loss": 0.5284,
+      "step": 376
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.150188323917138e-05,
+      "loss": 0.5098,
+      "step": 377
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.147834274952919e-05,
+      "loss": 0.4951,
+      "step": 378
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.145480225988701e-05,
+      "loss": 0.5022,
+      "step": 379
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.143126177024482e-05,
+      "loss": 0.4811,
+      "step": 380
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.140772128060264e-05,
+      "loss": 0.502,
+      "step": 381
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.138418079096045e-05,
+      "loss": 0.5423,
+      "step": 382
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.136064030131827e-05,
+      "loss": 0.4965,
+      "step": 383
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.133709981167608e-05,
+      "loss": 0.4984,
+      "step": 384
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.13135593220339e-05,
+      "loss": 0.5089,
+      "step": 385
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1290018832391714e-05,
+      "loss": 0.4733,
+      "step": 386
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.126647834274953e-05,
+      "loss": 0.5352,
+      "step": 387
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1242937853107345e-05,
+      "loss": 0.4944,
+      "step": 388
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1219397363465164e-05,
+      "loss": 0.4721,
+      "step": 389
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.119585687382298e-05,
+      "loss": 0.5265,
+      "step": 390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1172316384180795e-05,
+      "loss": 0.5017,
+      "step": 391
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1148775894538614e-05,
+      "loss": 0.5053,
+      "step": 392
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1125235404896426e-05,
+      "loss": 0.5089,
+      "step": 393
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.110169491525424e-05,
+      "loss": 0.4714,
+      "step": 394
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.107815442561205e-05,
+      "loss": 0.5145,
+      "step": 395
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.105461393596987e-05,
+      "loss": 0.4942,
+      "step": 396
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.103107344632768e-05,
+      "loss": 0.4956,
+      "step": 397
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.10075329566855e-05,
+      "loss": 0.4746,
+      "step": 398
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.098399246704331e-05,
+      "loss": 0.5216,
+      "step": 399
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.096045197740113e-05,
+      "loss": 0.5085,
+      "step": 400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.093691148775895e-05,
+      "loss": 0.5108,
+      "step": 401
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.091337099811676e-05,
+      "loss": 0.5026,
+      "step": 402
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.088983050847458e-05,
+      "loss": 0.4983,
+      "step": 403
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.086629001883239e-05,
+      "loss": 0.5148,
+      "step": 404
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.084274952919021e-05,
+      "loss": 0.4954,
+      "step": 405
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0819209039548024e-05,
+      "loss": 0.5079,
+      "step": 406
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.079566854990584e-05,
+      "loss": 0.5029,
+      "step": 407
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0772128060263655e-05,
+      "loss": 0.5049,
+      "step": 408
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0748587570621474e-05,
+      "loss": 0.5135,
+      "step": 409
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0725047080979286e-05,
+      "loss": 0.5032,
+      "step": 410
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.07015065913371e-05,
+      "loss": 0.4929,
+      "step": 411
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.067796610169492e-05,
+      "loss": 0.4984,
+      "step": 412
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.065442561205273e-05,
+      "loss": 0.4467,
+      "step": 413
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.063088512241055e-05,
+      "loss": 0.4978,
+      "step": 414
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.060734463276836e-05,
+      "loss": 0.4988,
+      "step": 415
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.058380414312618e-05,
+      "loss": 0.5107,
+      "step": 416
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.056026365348399e-05,
+      "loss": 0.478,
+      "step": 417
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.053672316384181e-05,
+      "loss": 0.5027,
+      "step": 418
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.051318267419962e-05,
+      "loss": 0.49,
+      "step": 419
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.048964218455744e-05,
+      "loss": 0.496,
+      "step": 420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0466101694915254e-05,
+      "loss": 0.5273,
+      "step": 421
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.044256120527307e-05,
+      "loss": 0.5034,
+      "step": 422
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.041902071563089e-05,
+      "loss": 0.4941,
+      "step": 423
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0395480225988704e-05,
+      "loss": 0.4966,
+      "step": 424
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.037193973634652e-05,
+      "loss": 0.5135,
+      "step": 425
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0348399246704335e-05,
+      "loss": 0.4765,
+      "step": 426
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0324858757062154e-05,
+      "loss": 0.4835,
+      "step": 427
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0301318267419966e-05,
+      "loss": 0.5303,
+      "step": 428
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.027777777777778e-05,
+      "loss": 0.4792,
+      "step": 429
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.025423728813559e-05,
+      "loss": 0.4759,
+      "step": 430
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.023069679849341e-05,
+      "loss": 0.5031,
+      "step": 431
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.020715630885122e-05,
+      "loss": 0.4768,
+      "step": 432
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.018361581920904e-05,
+      "loss": 0.5191,
+      "step": 433
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.016007532956685e-05,
+      "loss": 0.5168,
+      "step": 434
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.013653483992467e-05,
+      "loss": 0.5045,
+      "step": 435
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.011299435028249e-05,
+      "loss": 0.4893,
+      "step": 436
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.00894538606403e-05,
+      "loss": 0.5106,
+      "step": 437
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.006591337099812e-05,
+      "loss": 0.4979,
+      "step": 438
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.004237288135593e-05,
+      "loss": 0.4971,
+      "step": 439
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.001883239171375e-05,
+      "loss": 0.5009,
+      "step": 440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9995291902071564e-05,
+      "loss": 0.4828,
+      "step": 441
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.997175141242938e-05,
+      "loss": 0.5042,
+      "step": 442
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9948210922787195e-05,
+      "loss": 0.4693,
+      "step": 443
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9924670433145014e-05,
+      "loss": 0.5059,
+      "step": 444
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9901129943502826e-05,
+      "loss": 0.4799,
+      "step": 445
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9877589453860645e-05,
+      "loss": 0.4925,
+      "step": 446
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.985404896421846e-05,
+      "loss": 0.5136,
+      "step": 447
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.983050847457627e-05,
+      "loss": 0.5158,
+      "step": 448
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.980696798493409e-05,
+      "loss": 0.4987,
+      "step": 449
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.97834274952919e-05,
+      "loss": 0.5062,
+      "step": 450
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.975988700564972e-05,
+      "loss": 0.4867,
+      "step": 451
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.973634651600753e-05,
+      "loss": 0.4818,
+      "step": 452
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.971280602636535e-05,
+      "loss": 0.4997,
+      "step": 453
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.968926553672316e-05,
+      "loss": 0.5183,
+      "step": 454
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.966572504708098e-05,
+      "loss": 0.5085,
+      "step": 455
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9642184557438794e-05,
+      "loss": 0.4653,
+      "step": 456
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.961864406779661e-05,
+      "loss": 0.4889,
+      "step": 457
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.959510357815443e-05,
+      "loss": 0.5014,
+      "step": 458
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9571563088512244e-05,
+      "loss": 0.4944,
+      "step": 459
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.954802259887006e-05,
+      "loss": 0.4581,
+      "step": 460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9524482109227875e-05,
+      "loss": 0.4639,
+      "step": 461
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9500941619585694e-05,
+      "loss": 0.4978,
+      "step": 462
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9477401129943506e-05,
+      "loss": 0.4863,
+      "step": 463
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.945386064030132e-05,
+      "loss": 0.5284,
+      "step": 464
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.943032015065913e-05,
+      "loss": 0.5222,
+      "step": 465
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.940677966101695e-05,
+      "loss": 0.4941,
+      "step": 466
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.938323917137476e-05,
+      "loss": 0.5144,
+      "step": 467
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.935969868173258e-05,
+      "loss": 0.4862,
+      "step": 468
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.93361581920904e-05,
+      "loss": 0.5112,
+      "step": 469
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.931261770244821e-05,
+      "loss": 0.5102,
+      "step": 470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.928907721280603e-05,
+      "loss": 0.4949,
+      "step": 471
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.926553672316384e-05,
+      "loss": 0.4962,
+      "step": 472
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.924199623352166e-05,
+      "loss": 0.4955,
+      "step": 473
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.921845574387947e-05,
+      "loss": 0.5209,
+      "step": 474
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.919491525423729e-05,
+      "loss": 0.4548,
+      "step": 475
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.9171374764595104e-05,
+      "loss": 0.4636,
+      "step": 476
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.914783427495292e-05,
+      "loss": 0.4937,
+      "step": 477
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.9124293785310735e-05,
+      "loss": 0.4697,
+      "step": 478
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.9100753295668554e-05,
+      "loss": 0.5012,
+      "step": 479
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.9077212806026367e-05,
+      "loss": 0.479,
+      "step": 480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.9053672316384185e-05,
+      "loss": 0.4812,
+      "step": 481
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.9030131826742e-05,
+      "loss": 0.4881,
+      "step": 482
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.900659133709981e-05,
+      "loss": 0.4795,
+      "step": 483
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.898305084745763e-05,
+      "loss": 0.5102,
+      "step": 484
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.895951035781544e-05,
+      "loss": 0.4931,
+      "step": 485
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.893596986817326e-05,
+      "loss": 0.493,
+      "step": 486
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.891242937853107e-05,
+      "loss": 0.4967,
+      "step": 487
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.888888888888889e-05,
+      "loss": 0.4835,
+      "step": 488
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.88653483992467e-05,
+      "loss": 0.5025,
+      "step": 489
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.884180790960452e-05,
+      "loss": 0.4932,
+      "step": 490
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.881826741996234e-05,
+      "loss": 0.4913,
+      "step": 491
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.879472693032015e-05,
+      "loss": 0.5071,
+      "step": 492
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.877118644067797e-05,
+      "loss": 0.4814,
+      "step": 493
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8747645951035784e-05,
+      "loss": 0.4699,
+      "step": 494
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.87241054613936e-05,
+      "loss": 0.5089,
+      "step": 495
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8700564971751415e-05,
+      "loss": 0.5141,
+      "step": 496
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8677024482109234e-05,
+      "loss": 0.5076,
+      "step": 497
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8653483992467046e-05,
+      "loss": 0.5067,
+      "step": 498
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.862994350282486e-05,
+      "loss": 0.4829,
+      "step": 499
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.860640301318268e-05,
+      "loss": 0.4937,
+      "step": 500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.858286252354049e-05,
+      "loss": 0.4787,
+      "step": 501
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.855932203389831e-05,
+      "loss": 0.5082,
+      "step": 502
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.853578154425612e-05,
+      "loss": 0.4839,
+      "step": 503
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.851224105461394e-05,
+      "loss": 0.4693,
+      "step": 504
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.848870056497175e-05,
+      "loss": 0.4968,
+      "step": 505
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.846516007532957e-05,
+      "loss": 0.4651,
+      "step": 506
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.844161958568738e-05,
+      "loss": 0.5348,
+      "step": 507
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.84180790960452e-05,
+      "loss": 0.4914,
+      "step": 508
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8394538606403013e-05,
+      "loss": 0.5083,
+      "step": 509
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.837099811676083e-05,
+      "loss": 0.4989,
+      "step": 510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8347457627118644e-05,
+      "loss": 0.4678,
+      "step": 511
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8323917137476463e-05,
+      "loss": 0.494,
+      "step": 512
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8300376647834276e-05,
+      "loss": 0.4813,
+      "step": 513
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8276836158192094e-05,
+      "loss": 0.4862,
+      "step": 514
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8253295668549907e-05,
+      "loss": 0.4832,
+      "step": 515
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8229755178907726e-05,
+      "loss": 0.5081,
+      "step": 516
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.820621468926554e-05,
+      "loss": 0.4814,
+      "step": 517
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.818267419962335e-05,
+      "loss": 0.4842,
+      "step": 518
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.815913370998117e-05,
+      "loss": 0.498,
+      "step": 519
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.813559322033898e-05,
+      "loss": 0.4697,
+      "step": 520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.81120527306968e-05,
+      "loss": 0.4567,
+      "step": 521
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.808851224105461e-05,
+      "loss": 0.473,
+      "step": 522
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.806497175141243e-05,
+      "loss": 0.5038,
+      "step": 523
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.804143126177024e-05,
+      "loss": 0.4834,
+      "step": 524
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.801789077212806e-05,
+      "loss": 0.5165,
+      "step": 525
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.799435028248588e-05,
+      "loss": 0.4879,
+      "step": 526
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.797080979284369e-05,
+      "loss": 0.4752,
+      "step": 527
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.794726930320151e-05,
+      "loss": 0.5098,
+      "step": 528
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7923728813559324e-05,
+      "loss": 0.4674,
+      "step": 529
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.790018832391714e-05,
+      "loss": 0.5019,
+      "step": 530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7876647834274955e-05,
+      "loss": 0.4724,
+      "step": 531
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7853107344632774e-05,
+      "loss": 0.5085,
+      "step": 532
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7829566854990586e-05,
+      "loss": 0.4974,
+      "step": 533
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.78060263653484e-05,
+      "loss": 0.4865,
+      "step": 534
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.778248587570622e-05,
+      "loss": 0.4814,
+      "step": 535
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.775894538606403e-05,
+      "loss": 0.4886,
+      "step": 536
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.773540489642185e-05,
+      "loss": 0.4787,
+      "step": 537
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.771186440677966e-05,
+      "loss": 0.4762,
+      "step": 538
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.768832391713748e-05,
+      "loss": 0.4573,
+      "step": 539
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.766478342749529e-05,
+      "loss": 0.5255,
+      "step": 540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.764124293785311e-05,
+      "loss": 0.5303,
+      "step": 541
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.761770244821092e-05,
+      "loss": 0.5019,
+      "step": 542
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.759416195856874e-05,
+      "loss": 0.4892,
+      "step": 543
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7570621468926554e-05,
+      "loss": 0.4764,
+      "step": 544
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.754708097928437e-05,
+      "loss": 0.4892,
+      "step": 545
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7523540489642185e-05,
+      "loss": 0.5001,
+      "step": 546
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.4787,
+      "step": 547
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.747645951035782e-05,
+      "loss": 0.5032,
+      "step": 548
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7452919020715635e-05,
+      "loss": 0.4744,
+      "step": 549
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7429378531073453e-05,
+      "loss": 0.507,
+      "step": 550
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7405838041431266e-05,
+      "loss": 0.4858,
+      "step": 551
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.738229755178908e-05,
+      "loss": 0.5069,
+      "step": 552
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.735875706214689e-05,
+      "loss": 0.4967,
+      "step": 553
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.733521657250471e-05,
+      "loss": 0.4757,
+      "step": 554
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.731167608286252e-05,
+      "loss": 0.5101,
+      "step": 555
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.728813559322034e-05,
+      "loss": 0.476,
+      "step": 556
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.726459510357815e-05,
+      "loss": 0.4926,
+      "step": 557
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.724105461393597e-05,
+      "loss": 0.4604,
+      "step": 558
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.721751412429379e-05,
+      "loss": 0.4811,
+      "step": 559
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.71939736346516e-05,
+      "loss": 0.4581,
+      "step": 560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.717043314500942e-05,
+      "loss": 0.4958,
+      "step": 561
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.714689265536723e-05,
+      "loss": 0.4958,
+      "step": 562
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.712335216572505e-05,
+      "loss": 0.495,
+      "step": 563
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.7099811676082864e-05,
+      "loss": 0.4443,
+      "step": 564
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.707627118644068e-05,
+      "loss": 0.4925,
+      "step": 565
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.7052730696798495e-05,
+      "loss": 0.4963,
+      "step": 566
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.7029190207156314e-05,
+      "loss": 0.4874,
+      "step": 567
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7005649717514126e-05,
+      "loss": 0.4754,
+      "step": 568
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.698210922787194e-05,
+      "loss": 0.4912,
+      "step": 569
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.695856873822976e-05,
+      "loss": 0.4935,
+      "step": 570
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.693502824858757e-05,
+      "loss": 0.5005,
+      "step": 571
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.691148775894539e-05,
+      "loss": 0.4763,
+      "step": 572
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.68879472693032e-05,
+      "loss": 0.4925,
+      "step": 573
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.686440677966102e-05,
+      "loss": 0.4781,
+      "step": 574
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.684086629001883e-05,
+      "loss": 0.4908,
+      "step": 575
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.681732580037665e-05,
+      "loss": 0.4859,
+      "step": 576
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.679378531073446e-05,
+      "loss": 0.462,
+      "step": 577
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.677024482109228e-05,
+      "loss": 0.4967,
+      "step": 578
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6746704331450094e-05,
+      "loss": 0.502,
+      "step": 579
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.672316384180791e-05,
+      "loss": 0.4836,
+      "step": 580
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.669962335216573e-05,
+      "loss": 0.4904,
+      "step": 581
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6676082862523544e-05,
+      "loss": 0.4987,
+      "step": 582
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.665254237288136e-05,
+      "loss": 0.5051,
+      "step": 583
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6629001883239175e-05,
+      "loss": 0.4589,
+      "step": 584
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6605461393596994e-05,
+      "loss": 0.4972,
+      "step": 585
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6581920903954806e-05,
+      "loss": 0.495,
+      "step": 586
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.655838041431262e-05,
+      "loss": 0.4794,
+      "step": 587
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.653483992467043e-05,
+      "loss": 0.5103,
+      "step": 588
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.651129943502825e-05,
+      "loss": 0.4753,
+      "step": 589
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.648775894538606e-05,
+      "loss": 0.4925,
+      "step": 590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.646421845574388e-05,
+      "loss": 0.4866,
+      "step": 591
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.644067796610169e-05,
+      "loss": 0.4986,
+      "step": 592
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.641713747645951e-05,
+      "loss": 0.5242,
+      "step": 593
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.639359698681733e-05,
+      "loss": 0.4947,
+      "step": 594
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.637005649717514e-05,
+      "loss": 0.4604,
+      "step": 595
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.634651600753296e-05,
+      "loss": 0.4959,
+      "step": 596
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.632297551789077e-05,
+      "loss": 0.5129,
+      "step": 597
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.629943502824859e-05,
+      "loss": 0.4741,
+      "step": 598
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6275894538606404e-05,
+      "loss": 0.4733,
+      "step": 599
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.625235404896422e-05,
+      "loss": 0.4838,
+      "step": 600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6228813559322035e-05,
+      "loss": 0.5002,
+      "step": 601
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6205273069679854e-05,
+      "loss": 0.4974,
+      "step": 602
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6181732580037666e-05,
+      "loss": 0.4639,
+      "step": 603
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.6158192090395485e-05,
+      "loss": 0.5054,
+      "step": 604
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.61346516007533e-05,
+      "loss": 0.4973,
+      "step": 605
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.611111111111111e-05,
+      "loss": 0.4724,
+      "step": 606
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.608757062146893e-05,
+      "loss": 0.461,
+      "step": 607
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.606403013182674e-05,
+      "loss": 0.4632,
+      "step": 608
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.604048964218456e-05,
+      "loss": 0.49,
+      "step": 609
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.601694915254237e-05,
+      "loss": 0.4597,
+      "step": 610
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.599340866290019e-05,
+      "loss": 0.4964,
+      "step": 611
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5969868173258e-05,
+      "loss": 0.4942,
+      "step": 612
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.594632768361582e-05,
+      "loss": 0.489,
+      "step": 613
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5922787193973634e-05,
+      "loss": 0.4762,
+      "step": 614
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.589924670433145e-05,
+      "loss": 0.4757,
+      "step": 615
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.587570621468927e-05,
+      "loss": 0.4725,
+      "step": 616
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5852165725047084e-05,
+      "loss": 0.4845,
+      "step": 617
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.58286252354049e-05,
+      "loss": 0.4646,
+      "step": 618
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5805084745762715e-05,
+      "loss": 0.4916,
+      "step": 619
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5781544256120534e-05,
+      "loss": 0.4502,
+      "step": 620
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5758003766478346e-05,
+      "loss": 0.4653,
+      "step": 621
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.573446327683616e-05,
+      "loss": 0.4503,
+      "step": 622
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.571092278719397e-05,
+      "loss": 0.4364,
+      "step": 623
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.568738229755179e-05,
+      "loss": 0.479,
+      "step": 624
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.56638418079096e-05,
+      "loss": 0.4447,
+      "step": 625
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.564030131826742e-05,
+      "loss": 0.4467,
+      "step": 626
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.561676082862524e-05,
+      "loss": 0.4746,
+      "step": 627
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.559322033898305e-05,
+      "loss": 0.4878,
+      "step": 628
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.556967984934087e-05,
+      "loss": 0.4678,
+      "step": 629
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.554613935969868e-05,
+      "loss": 0.4733,
+      "step": 630
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.55225988700565e-05,
+      "loss": 0.4776,
+      "step": 631
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.549905838041431e-05,
+      "loss": 0.4444,
+      "step": 632
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.547551789077213e-05,
+      "loss": 0.4893,
+      "step": 633
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5451977401129944e-05,
+      "loss": 0.4631,
+      "step": 634
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.542843691148776e-05,
+      "loss": 0.465,
+      "step": 635
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5404896421845575e-05,
+      "loss": 0.4779,
+      "step": 636
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5381355932203394e-05,
+      "loss": 0.4981,
+      "step": 637
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5357815442561206e-05,
+      "loss": 0.476,
+      "step": 638
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5334274952919025e-05,
+      "loss": 0.4911,
+      "step": 639
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.531073446327684e-05,
+      "loss": 0.4853,
+      "step": 640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.528719397363465e-05,
+      "loss": 0.5485,
+      "step": 641
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.526365348399247e-05,
+      "loss": 0.4812,
+      "step": 642
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.524011299435028e-05,
+      "loss": 0.4705,
+      "step": 643
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.52165725047081e-05,
+      "loss": 0.4978,
+      "step": 644
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.519303201506591e-05,
+      "loss": 0.4954,
+      "step": 645
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.516949152542373e-05,
+      "loss": 0.4985,
+      "step": 646
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.514595103578154e-05,
+      "loss": 0.4669,
+      "step": 647
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.512241054613936e-05,
+      "loss": 0.4824,
+      "step": 648
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.509887005649718e-05,
+      "loss": 0.4583,
+      "step": 649
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.507532956685499e-05,
+      "loss": 0.4939,
+      "step": 650
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.505178907721281e-05,
+      "loss": 0.4826,
+      "step": 651
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5028248587570624e-05,
+      "loss": 0.4674,
+      "step": 652
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.500470809792844e-05,
+      "loss": 0.4907,
+      "step": 653
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4981167608286255e-05,
+      "loss": 0.4851,
+      "step": 654
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4957627118644074e-05,
+      "loss": 0.4773,
+      "step": 655
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4934086629001886e-05,
+      "loss": 0.4807,
+      "step": 656
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.49105461393597e-05,
+      "loss": 0.4914,
+      "step": 657
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.488700564971752e-05,
+      "loss": 0.4783,
+      "step": 658
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.486346516007533e-05,
+      "loss": 0.4758,
+      "step": 659
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.483992467043315e-05,
+      "loss": 0.4721,
+      "step": 660
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.481638418079096e-05,
+      "loss": 0.4899,
+      "step": 661
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.479284369114878e-05,
+      "loss": 0.441,
+      "step": 662
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.476930320150659e-05,
+      "loss": 0.4512,
+      "step": 663
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.474576271186441e-05,
+      "loss": 0.4795,
+      "step": 664
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.472222222222222e-05,
+      "loss": 0.4682,
+      "step": 665
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.469868173258004e-05,
+      "loss": 0.4892,
+      "step": 666
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.467514124293785e-05,
+      "loss": 0.4772,
+      "step": 667
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.465160075329567e-05,
+      "loss": 0.5147,
+      "step": 668
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4628060263653484e-05,
+      "loss": 0.4851,
+      "step": 669
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.46045197740113e-05,
+      "loss": 0.4849,
+      "step": 670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4580979284369115e-05,
+      "loss": 0.4749,
+      "step": 671
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4557438794726934e-05,
+      "loss": 0.4728,
+      "step": 672
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4533898305084746e-05,
+      "loss": 0.4578,
+      "step": 673
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4510357815442565e-05,
+      "loss": 0.4695,
+      "step": 674
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.448681732580038e-05,
+      "loss": 0.4981,
+      "step": 675
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.446327683615819e-05,
+      "loss": 0.454,
+      "step": 676
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.443973634651601e-05,
+      "loss": 0.4892,
+      "step": 677
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.441619585687382e-05,
+      "loss": 0.4615,
+      "step": 678
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.439265536723164e-05,
+      "loss": 0.4782,
+      "step": 679
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.436911487758945e-05,
+      "loss": 0.4744,
+      "step": 680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.434557438794727e-05,
+      "loss": 0.4624,
+      "step": 681
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.432203389830508e-05,
+      "loss": 0.4844,
+      "step": 682
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.42984934086629e-05,
+      "loss": 0.4867,
+      "step": 683
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.427495291902072e-05,
+      "loss": 0.478,
+      "step": 684
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.425141242937853e-05,
+      "loss": 0.4722,
+      "step": 685
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.422787193973635e-05,
+      "loss": 0.4767,
+      "step": 686
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4204331450094164e-05,
+      "loss": 0.4851,
+      "step": 687
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.418079096045198e-05,
+      "loss": 0.4741,
+      "step": 688
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4157250470809795e-05,
+      "loss": 0.4597,
+      "step": 689
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4133709981167614e-05,
+      "loss": 0.4563,
+      "step": 690
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4110169491525426e-05,
+      "loss": 0.4525,
+      "step": 691
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.408662900188324e-05,
+      "loss": 0.4748,
+      "step": 692
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.406308851224106e-05,
+      "loss": 0.471,
+      "step": 693
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.403954802259887e-05,
+      "loss": 0.5225,
+      "step": 694
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.401600753295669e-05,
+      "loss": 0.4795,
+      "step": 695
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.39924670433145e-05,
+      "loss": 0.4661,
+      "step": 696
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.396892655367232e-05,
+      "loss": 0.4867,
+      "step": 697
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.394538606403013e-05,
+      "loss": 0.4758,
+      "step": 698
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.392184557438795e-05,
+      "loss": 0.5136,
+      "step": 699
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.389830508474576e-05,
+      "loss": 0.4526,
+      "step": 700
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.387476459510358e-05,
+      "loss": 0.4766,
+      "step": 701
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.385122410546139e-05,
+      "loss": 0.5078,
+      "step": 702
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.382768361581921e-05,
+      "loss": 0.4763,
+      "step": 703
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3804143126177024e-05,
+      "loss": 0.4586,
+      "step": 704
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.378060263653484e-05,
+      "loss": 0.4808,
+      "step": 705
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.375706214689266e-05,
+      "loss": 0.4579,
+      "step": 706
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3733521657250474e-05,
+      "loss": 0.5064,
+      "step": 707
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.370998116760829e-05,
+      "loss": 0.5042,
+      "step": 708
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3686440677966105e-05,
+      "loss": 0.5036,
+      "step": 709
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.366290018832392e-05,
+      "loss": 0.4683,
+      "step": 710
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.363935969868173e-05,
+      "loss": 0.4372,
+      "step": 711
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.361581920903955e-05,
+      "loss": 0.4931,
+      "step": 712
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.359227871939736e-05,
+      "loss": 0.4734,
+      "step": 713
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.356873822975518e-05,
+      "loss": 0.4953,
+      "step": 714
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.354519774011299e-05,
+      "loss": 0.4088,
+      "step": 715
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.352165725047081e-05,
+      "loss": 0.3867,
+      "step": 716
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.349811676082863e-05,
+      "loss": 0.4031,
+      "step": 717
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.347457627118644e-05,
+      "loss": 0.3789,
+      "step": 718
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.345103578154426e-05,
+      "loss": 0.3895,
+      "step": 719
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.342749529190207e-05,
+      "loss": 0.3895,
+      "step": 720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.340395480225989e-05,
+      "loss": 0.406,
+      "step": 721
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3380414312617704e-05,
+      "loss": 0.3988,
+      "step": 722
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.335687382297552e-05,
+      "loss": 0.3782,
+      "step": 723
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.396,
+      "step": 724
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3309792843691154e-05,
+      "loss": 0.3931,
+      "step": 725
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3286252354048966e-05,
+      "loss": 0.3819,
+      "step": 726
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.326271186440678e-05,
+      "loss": 0.4045,
+      "step": 727
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.32391713747646e-05,
+      "loss": 0.3863,
+      "step": 728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.321563088512241e-05,
+      "loss": 0.3899,
+      "step": 729
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.319209039548023e-05,
+      "loss": 0.3741,
+      "step": 730
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.316854990583804e-05,
+      "loss": 0.3797,
+      "step": 731
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.314500941619586e-05,
+      "loss": 0.364,
+      "step": 732
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.312146892655367e-05,
+      "loss": 0.3673,
+      "step": 733
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.309792843691149e-05,
+      "loss": 0.3615,
+      "step": 734
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.30743879472693e-05,
+      "loss": 0.3832,
+      "step": 735
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.305084745762712e-05,
+      "loss": 0.394,
+      "step": 736
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.3027306967984933e-05,
+      "loss": 0.3756,
+      "step": 737
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.300376647834275e-05,
+      "loss": 0.3771,
+      "step": 738
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2980225988700564e-05,
+      "loss": 0.3706,
+      "step": 739
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.295668549905838e-05,
+      "loss": 0.3853,
+      "step": 740
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.29331450094162e-05,
+      "loss": 0.3966,
+      "step": 741
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2909604519774014e-05,
+      "loss": 0.3781,
+      "step": 742
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.288606403013183e-05,
+      "loss": 0.3727,
+      "step": 743
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2862523540489645e-05,
+      "loss": 0.3844,
+      "step": 744
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.283898305084746e-05,
+      "loss": 0.3916,
+      "step": 745
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.281544256120527e-05,
+      "loss": 0.3593,
+      "step": 746
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.279190207156309e-05,
+      "loss": 0.4072,
+      "step": 747
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.27683615819209e-05,
+      "loss": 0.3773,
+      "step": 748
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.274482109227872e-05,
+      "loss": 0.383,
+      "step": 749
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.272128060263653e-05,
+      "loss": 0.3967,
+      "step": 750
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.269774011299435e-05,
+      "loss": 0.3686,
+      "step": 751
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.267419962335217e-05,
+      "loss": 0.3727,
+      "step": 752
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.265065913370998e-05,
+      "loss": 0.3873,
+      "step": 753
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.26271186440678e-05,
+      "loss": 0.3921,
+      "step": 754
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.260357815442561e-05,
+      "loss": 0.3583,
+      "step": 755
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.258003766478343e-05,
+      "loss": 0.3706,
+      "step": 756
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2556497175141244e-05,
+      "loss": 0.3644,
+      "step": 757
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.253295668549906e-05,
+      "loss": 0.3684,
+      "step": 758
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2509416195856875e-05,
+      "loss": 0.3911,
+      "step": 759
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2485875706214694e-05,
+      "loss": 0.3886,
+      "step": 760
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2462335216572506e-05,
+      "loss": 0.3777,
+      "step": 761
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2438794726930325e-05,
+      "loss": 0.3661,
+      "step": 762
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.241525423728814e-05,
+      "loss": 0.3605,
+      "step": 763
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.239171374764595e-05,
+      "loss": 0.3752,
+      "step": 764
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.236817325800377e-05,
+      "loss": 0.3758,
+      "step": 765
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.234463276836158e-05,
+      "loss": 0.3673,
+      "step": 766
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.23210922787194e-05,
+      "loss": 0.3677,
+      "step": 767
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.229755178907721e-05,
+      "loss": 0.3622,
+      "step": 768
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.227401129943503e-05,
+      "loss": 0.3708,
+      "step": 769
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.225047080979284e-05,
+      "loss": 0.3867,
+      "step": 770
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.222693032015066e-05,
+      "loss": 0.3747,
+      "step": 771
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2203389830508473e-05,
+      "loss": 0.3965,
+      "step": 772
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.217984934086629e-05,
+      "loss": 0.383,
+      "step": 773
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.215630885122411e-05,
+      "loss": 0.3996,
+      "step": 774
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.2132768361581923e-05,
+      "loss": 0.3706,
+      "step": 775
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.210922787193974e-05,
+      "loss": 0.3824,
+      "step": 776
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.2085687382297555e-05,
+      "loss": 0.3861,
+      "step": 777
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.2062146892655373e-05,
+      "loss": 0.4104,
+      "step": 778
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.2038606403013186e-05,
+      "loss": 0.3816,
+      "step": 779
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.2015065913371e-05,
+      "loss": 0.3776,
+      "step": 780
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.199152542372881e-05,
+      "loss": 0.4013,
+      "step": 781
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.196798493408663e-05,
+      "loss": 0.3705,
+      "step": 782
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.194444444444444e-05,
+      "loss": 0.3732,
+      "step": 783
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.192090395480226e-05,
+      "loss": 0.3718,
+      "step": 784
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.189736346516008e-05,
+      "loss": 0.3833,
+      "step": 785
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.187382297551789e-05,
+      "loss": 0.3793,
+      "step": 786
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.185028248587571e-05,
+      "loss": 0.3636,
+      "step": 787
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.182674199623352e-05,
+      "loss": 0.4029,
+      "step": 788
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.180320150659134e-05,
+      "loss": 0.3868,
+      "step": 789
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.177966101694915e-05,
+      "loss": 0.3989,
+      "step": 790
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.175612052730697e-05,
+      "loss": 0.3642,
+      "step": 791
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1732580037664784e-05,
+      "loss": 0.3811,
+      "step": 792
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.17090395480226e-05,
+      "loss": 0.3862,
+      "step": 793
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1685499058380415e-05,
+      "loss": 0.3742,
+      "step": 794
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1661958568738234e-05,
+      "loss": 0.3945,
+      "step": 795
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1638418079096046e-05,
+      "loss": 0.3815,
+      "step": 796
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1614877589453865e-05,
+      "loss": 0.3918,
+      "step": 797
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.159133709981168e-05,
+      "loss": 0.401,
+      "step": 798
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.156779661016949e-05,
+      "loss": 0.3874,
+      "step": 799
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.154425612052731e-05,
+      "loss": 0.3787,
+      "step": 800
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.152071563088512e-05,
+      "loss": 0.3661,
+      "step": 801
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.149717514124294e-05,
+      "loss": 0.3849,
+      "step": 802
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.147363465160075e-05,
+      "loss": 0.3874,
+      "step": 803
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.145009416195857e-05,
+      "loss": 0.3813,
+      "step": 804
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.142655367231638e-05,
+      "loss": 0.4047,
+      "step": 805
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.14030131826742e-05,
+      "loss": 0.3811,
+      "step": 806
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.137947269303202e-05,
+      "loss": 0.3522,
+      "step": 807
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.135593220338983e-05,
+      "loss": 0.366,
+      "step": 808
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.133239171374765e-05,
+      "loss": 0.3921,
+      "step": 809
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1308851224105464e-05,
+      "loss": 0.3841,
+      "step": 810
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.128531073446328e-05,
+      "loss": 0.3777,
+      "step": 811
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1261770244821095e-05,
+      "loss": 0.3702,
+      "step": 812
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1238229755178913e-05,
+      "loss": 0.384,
+      "step": 813
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1214689265536726e-05,
+      "loss": 0.3803,
+      "step": 814
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.119114877589454e-05,
+      "loss": 0.3816,
+      "step": 815
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.116760828625235e-05,
+      "loss": 0.3689,
+      "step": 816
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.114406779661017e-05,
+      "loss": 0.3723,
+      "step": 817
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.112052730696798e-05,
+      "loss": 0.3762,
+      "step": 818
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.10969868173258e-05,
+      "loss": 0.3859,
+      "step": 819
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.107344632768362e-05,
+      "loss": 0.3959,
+      "step": 820
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.104990583804143e-05,
+      "loss": 0.3772,
+      "step": 821
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.102636534839925e-05,
+      "loss": 0.407,
+      "step": 822
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.100282485875706e-05,
+      "loss": 0.3984,
+      "step": 823
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.097928436911488e-05,
+      "loss": 0.3748,
+      "step": 824
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.095574387947269e-05,
+      "loss": 0.3886,
+      "step": 825
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.093220338983051e-05,
+      "loss": 0.3697,
+      "step": 826
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0908662900188324e-05,
+      "loss": 0.3883,
+      "step": 827
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.088512241054614e-05,
+      "loss": 0.3692,
+      "step": 828
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0861581920903955e-05,
+      "loss": 0.3724,
+      "step": 829
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0838041431261774e-05,
+      "loss": 0.3782,
+      "step": 830
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0814500941619586e-05,
+      "loss": 0.3597,
+      "step": 831
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0790960451977405e-05,
+      "loss": 0.3671,
+      "step": 832
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.076741996233522e-05,
+      "loss": 0.3697,
+      "step": 833
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.074387947269303e-05,
+      "loss": 0.3607,
+      "step": 834
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.072033898305085e-05,
+      "loss": 0.3753,
+      "step": 835
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.069679849340866e-05,
+      "loss": 0.3843,
+      "step": 836
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.067325800376648e-05,
+      "loss": 0.3921,
+      "step": 837
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.064971751412429e-05,
+      "loss": 0.3699,
+      "step": 838
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.062617702448211e-05,
+      "loss": 0.3533,
+      "step": 839
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.060263653483992e-05,
+      "loss": 0.3806,
+      "step": 840
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.057909604519774e-05,
+      "loss": 0.3755,
+      "step": 841
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.055555555555556e-05,
+      "loss": 0.393,
+      "step": 842
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.053201506591337e-05,
+      "loss": 0.3808,
+      "step": 843
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.050847457627119e-05,
+      "loss": 0.3934,
+      "step": 844
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0484934086629004e-05,
+      "loss": 0.3767,
+      "step": 845
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.046139359698682e-05,
+      "loss": 0.3584,
+      "step": 846
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.043785310734463e-05,
+      "loss": 0.3736,
+      "step": 847
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.041431261770245e-05,
+      "loss": 0.399,
+      "step": 848
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0390772128060262e-05,
+      "loss": 0.3773,
+      "step": 849
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.036723163841808e-05,
+      "loss": 0.3876,
+      "step": 850
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0343691148775893e-05,
+      "loss": 0.3878,
+      "step": 851
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0320150659133712e-05,
+      "loss": 0.3808,
+      "step": 852
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0296610169491528e-05,
+      "loss": 0.3642,
+      "step": 853
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0273069679849343e-05,
+      "loss": 0.3807,
+      "step": 854
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.024952919020716e-05,
+      "loss": 0.3713,
+      "step": 855
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.022598870056497e-05,
+      "loss": 0.3728,
+      "step": 856
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.020244821092279e-05,
+      "loss": 0.384,
+      "step": 857
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0178907721280602e-05,
+      "loss": 0.3842,
+      "step": 858
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.015536723163842e-05,
+      "loss": 0.362,
+      "step": 859
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0131826741996233e-05,
+      "loss": 0.3709,
+      "step": 860
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.0108286252354052e-05,
+      "loss": 0.3743,
+      "step": 861
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.0084745762711864e-05,
+      "loss": 0.3809,
+      "step": 862
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.006120527306968e-05,
+      "loss": 0.3809,
+      "step": 863
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.00376647834275e-05,
+      "loss": 0.3607,
+      "step": 864
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.001412429378531e-05,
+      "loss": 0.3711,
+      "step": 865
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.999058380414313e-05,
+      "loss": 0.3662,
+      "step": 866
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9967043314500942e-05,
+      "loss": 0.3569,
+      "step": 867
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.994350282485876e-05,
+      "loss": 0.3818,
+      "step": 868
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9919962335216573e-05,
+      "loss": 0.3648,
+      "step": 869
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9896421845574392e-05,
+      "loss": 0.369,
+      "step": 870
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9872881355932204e-05,
+      "loss": 0.3726,
+      "step": 871
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.984934086629002e-05,
+      "loss": 0.3946,
+      "step": 872
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9825800376647835e-05,
+      "loss": 0.3764,
+      "step": 873
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.980225988700565e-05,
+      "loss": 0.3831,
+      "step": 874
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.977871939736347e-05,
+      "loss": 0.3796,
+      "step": 875
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.975517890772128e-05,
+      "loss": 0.3783,
+      "step": 876
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.97316384180791e-05,
+      "loss": 0.3713,
+      "step": 877
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9708097928436913e-05,
+      "loss": 0.37,
+      "step": 878
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.968455743879473e-05,
+      "loss": 0.3863,
+      "step": 879
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9661016949152544e-05,
+      "loss": 0.3727,
+      "step": 880
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.963747645951036e-05,
+      "loss": 0.3694,
+      "step": 881
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.961393596986817e-05,
+      "loss": 0.3904,
+      "step": 882
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.959039548022599e-05,
+      "loss": 0.3843,
+      "step": 883
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9566854990583802e-05,
+      "loss": 0.3846,
+      "step": 884
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.954331450094162e-05,
+      "loss": 0.3798,
+      "step": 885
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.951977401129944e-05,
+      "loss": 0.3759,
+      "step": 886
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9496233521657252e-05,
+      "loss": 0.398,
+      "step": 887
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.947269303201507e-05,
+      "loss": 0.3732,
+      "step": 888
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9449152542372883e-05,
+      "loss": 0.3799,
+      "step": 889
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.94256120527307e-05,
+      "loss": 0.3855,
+      "step": 890
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.940207156308851e-05,
+      "loss": 0.3748,
+      "step": 891
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.937853107344633e-05,
+      "loss": 0.382,
+      "step": 892
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9354990583804142e-05,
+      "loss": 0.374,
+      "step": 893
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.933145009416196e-05,
+      "loss": 0.3874,
+      "step": 894
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9307909604519773e-05,
+      "loss": 0.3763,
+      "step": 895
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9284369114877592e-05,
+      "loss": 0.3658,
+      "step": 896
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9260828625235404e-05,
+      "loss": 0.3994,
+      "step": 897
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9237288135593223e-05,
+      "loss": 0.382,
+      "step": 898
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.921374764595104e-05,
+      "loss": 0.3727,
+      "step": 899
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.919020715630885e-05,
+      "loss": 0.3945,
+      "step": 900
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.916666666666667e-05,
+      "loss": 0.391,
+      "step": 901
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9143126177024482e-05,
+      "loss": 0.3807,
+      "step": 902
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.91195856873823e-05,
+      "loss": 0.3881,
+      "step": 903
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.9096045197740113e-05,
+      "loss": 0.3981,
+      "step": 904
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.9072504708097932e-05,
+      "loss": 0.3616,
+      "step": 905
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.9048964218455744e-05,
+      "loss": 0.3825,
+      "step": 906
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.902542372881356e-05,
+      "loss": 0.3713,
+      "step": 907
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.9001883239171375e-05,
+      "loss": 0.3877,
+      "step": 908
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.897834274952919e-05,
+      "loss": 0.357,
+      "step": 909
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.895480225988701e-05,
+      "loss": 0.3896,
+      "step": 910
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.893126177024482e-05,
+      "loss": 0.385,
+      "step": 911
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.890772128060264e-05,
+      "loss": 0.3621,
+      "step": 912
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8884180790960453e-05,
+      "loss": 0.3766,
+      "step": 913
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.886064030131827e-05,
+      "loss": 0.3915,
+      "step": 914
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8837099811676084e-05,
+      "loss": 0.3845,
+      "step": 915
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.88135593220339e-05,
+      "loss": 0.3784,
+      "step": 916
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.879001883239171e-05,
+      "loss": 0.3753,
+      "step": 917
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.876647834274953e-05,
+      "loss": 0.3767,
+      "step": 918
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8742937853107342e-05,
+      "loss": 0.3891,
+      "step": 919
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.871939736346516e-05,
+      "loss": 0.3864,
+      "step": 920
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.869585687382298e-05,
+      "loss": 0.3598,
+      "step": 921
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8672316384180792e-05,
+      "loss": 0.3824,
+      "step": 922
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.864877589453861e-05,
+      "loss": 0.3705,
+      "step": 923
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8625235404896424e-05,
+      "loss": 0.3845,
+      "step": 924
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.860169491525424e-05,
+      "loss": 0.362,
+      "step": 925
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.857815442561205e-05,
+      "loss": 0.3912,
+      "step": 926
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.855461393596987e-05,
+      "loss": 0.3819,
+      "step": 927
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8531073446327682e-05,
+      "loss": 0.3888,
+      "step": 928
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.85075329566855e-05,
+      "loss": 0.3829,
+      "step": 929
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8483992467043313e-05,
+      "loss": 0.3812,
+      "step": 930
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8460451977401132e-05,
+      "loss": 0.3906,
+      "step": 931
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8436911487758948e-05,
+      "loss": 0.3695,
+      "step": 932
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8413370998116763e-05,
+      "loss": 0.3817,
+      "step": 933
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.838983050847458e-05,
+      "loss": 0.373,
+      "step": 934
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.836629001883239e-05,
+      "loss": 0.3738,
+      "step": 935
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.834274952919021e-05,
+      "loss": 0.3788,
+      "step": 936
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8319209039548022e-05,
+      "loss": 0.4016,
+      "step": 937
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.829566854990584e-05,
+      "loss": 0.4062,
+      "step": 938
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8272128060263653e-05,
+      "loss": 0.3787,
+      "step": 939
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8248587570621472e-05,
+      "loss": 0.3883,
+      "step": 940
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8225047080979284e-05,
+      "loss": 0.3689,
+      "step": 941
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.82015065913371e-05,
+      "loss": 0.3532,
+      "step": 942
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.817796610169492e-05,
+      "loss": 0.3964,
+      "step": 943
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.815442561205273e-05,
+      "loss": 0.3768,
+      "step": 944
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.813088512241055e-05,
+      "loss": 0.3845,
+      "step": 945
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.8107344632768362e-05,
+      "loss": 0.3776,
+      "step": 946
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.808380414312618e-05,
+      "loss": 0.392,
+      "step": 947
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.8060263653483993e-05,
+      "loss": 0.3668,
+      "step": 948
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.803672316384181e-05,
+      "loss": 0.3849,
+      "step": 949
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.8013182674199624e-05,
+      "loss": 0.3486,
+      "step": 950
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.798964218455744e-05,
+      "loss": 0.3734,
+      "step": 951
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7966101694915255e-05,
+      "loss": 0.4048,
+      "step": 952
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.794256120527307e-05,
+      "loss": 0.3804,
+      "step": 953
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.791902071563089e-05,
+      "loss": 0.3728,
+      "step": 954
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.78954802259887e-05,
+      "loss": 0.375,
+      "step": 955
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.787193973634652e-05,
+      "loss": 0.3753,
+      "step": 956
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7848399246704333e-05,
+      "loss": 0.3716,
+      "step": 957
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.782485875706215e-05,
+      "loss": 0.3535,
+      "step": 958
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7801318267419964e-05,
+      "loss": 0.3786,
+      "step": 959
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.777777777777778e-05,
+      "loss": 0.3804,
+      "step": 960
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.775423728813559e-05,
+      "loss": 0.3909,
+      "step": 961
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.773069679849341e-05,
+      "loss": 0.386,
+      "step": 962
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7707156308851222e-05,
+      "loss": 0.3661,
+      "step": 963
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.768361581920904e-05,
+      "loss": 0.3837,
+      "step": 964
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7660075329566853e-05,
+      "loss": 0.3782,
+      "step": 965
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7636534839924672e-05,
+      "loss": 0.3754,
+      "step": 966
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7612994350282488e-05,
+      "loss": 0.3893,
+      "step": 967
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7589453860640303e-05,
+      "loss": 0.3792,
+      "step": 968
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.756591337099812e-05,
+      "loss": 0.389,
+      "step": 969
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.754237288135593e-05,
+      "loss": 0.4018,
+      "step": 970
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.751883239171375e-05,
+      "loss": 0.3615,
+      "step": 971
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7495291902071562e-05,
+      "loss": 0.3864,
+      "step": 972
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.747175141242938e-05,
+      "loss": 0.3839,
+      "step": 973
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7448210922787193e-05,
+      "loss": 0.3807,
+      "step": 974
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7424670433145012e-05,
+      "loss": 0.366,
+      "step": 975
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7401129943502824e-05,
+      "loss": 0.3919,
+      "step": 976
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7377589453860643e-05,
+      "loss": 0.387,
+      "step": 977
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.735404896421846e-05,
+      "loss": 0.384,
+      "step": 978
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.733050847457627e-05,
+      "loss": 0.4012,
+      "step": 979
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.730696798493409e-05,
+      "loss": 0.385,
+      "step": 980
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7283427495291902e-05,
+      "loss": 0.3846,
+      "step": 981
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.725988700564972e-05,
+      "loss": 0.3574,
+      "step": 982
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7236346516007533e-05,
+      "loss": 0.4029,
+      "step": 983
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7212806026365352e-05,
+      "loss": 0.3748,
+      "step": 984
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7189265536723164e-05,
+      "loss": 0.3683,
+      "step": 985
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.716572504708098e-05,
+      "loss": 0.3596,
+      "step": 986
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7142184557438795e-05,
+      "loss": 0.4124,
+      "step": 987
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.711864406779661e-05,
+      "loss": 0.3668,
+      "step": 988
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.709510357815443e-05,
+      "loss": 0.3729,
+      "step": 989
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.707156308851224e-05,
+      "loss": 0.3872,
+      "step": 990
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.704802259887006e-05,
+      "loss": 0.3598,
+      "step": 991
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.7024482109227873e-05,
+      "loss": 0.3945,
+      "step": 992
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.700094161958569e-05,
+      "loss": 0.3626,
+      "step": 993
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6977401129943504e-05,
+      "loss": 0.3817,
+      "step": 994
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.695386064030132e-05,
+      "loss": 0.3976,
+      "step": 995
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.693032015065913e-05,
+      "loss": 0.3628,
+      "step": 996
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.690677966101695e-05,
+      "loss": 0.373,
+      "step": 997
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6883239171374762e-05,
+      "loss": 0.386,
+      "step": 998
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.685969868173258e-05,
+      "loss": 0.3832,
+      "step": 999
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.68361581920904e-05,
+      "loss": 0.3919,
+      "step": 1000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6812617702448212e-05,
+      "loss": 0.38,
+      "step": 1001
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.678907721280603e-05,
+      "loss": 0.3782,
+      "step": 1002
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6765536723163843e-05,
+      "loss": 0.3769,
+      "step": 1003
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.674199623352166e-05,
+      "loss": 0.3722,
+      "step": 1004
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.671845574387947e-05,
+      "loss": 0.3716,
+      "step": 1005
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.669491525423729e-05,
+      "loss": 0.3951,
+      "step": 1006
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6671374764595102e-05,
+      "loss": 0.3739,
+      "step": 1007
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.664783427495292e-05,
+      "loss": 0.3593,
+      "step": 1008
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6624293785310733e-05,
+      "loss": 0.3601,
+      "step": 1009
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6600753295668552e-05,
+      "loss": 0.3736,
+      "step": 1010
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6577212806026368e-05,
+      "loss": 0.3498,
+      "step": 1011
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6553672316384183e-05,
+      "loss": 0.366,
+      "step": 1012
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6530131826742e-05,
+      "loss": 0.3913,
+      "step": 1013
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.650659133709981e-05,
+      "loss": 0.3679,
+      "step": 1014
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.648305084745763e-05,
+      "loss": 0.3653,
+      "step": 1015
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6459510357815442e-05,
+      "loss": 0.39,
+      "step": 1016
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.643596986817326e-05,
+      "loss": 0.3818,
+      "step": 1017
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6412429378531073e-05,
+      "loss": 0.3911,
+      "step": 1018
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6388888888888892e-05,
+      "loss": 0.3758,
+      "step": 1019
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6365348399246704e-05,
+      "loss": 0.3674,
+      "step": 1020
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.634180790960452e-05,
+      "loss": 0.4023,
+      "step": 1021
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.631826741996234e-05,
+      "loss": 0.3954,
+      "step": 1022
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.629472693032015e-05,
+      "loss": 0.3976,
+      "step": 1023
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.627118644067797e-05,
+      "loss": 0.3679,
+      "step": 1024
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.624764595103578e-05,
+      "loss": 0.3706,
+      "step": 1025
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.62241054613936e-05,
+      "loss": 0.3883,
+      "step": 1026
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6200564971751413e-05,
+      "loss": 0.3699,
+      "step": 1027
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.617702448210923e-05,
+      "loss": 0.3824,
+      "step": 1028
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6153483992467044e-05,
+      "loss": 0.3865,
+      "step": 1029
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.612994350282486e-05,
+      "loss": 0.3588,
+      "step": 1030
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6106403013182675e-05,
+      "loss": 0.3894,
+      "step": 1031
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.608286252354049e-05,
+      "loss": 0.4032,
+      "step": 1032
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.605932203389831e-05,
+      "loss": 0.37,
+      "step": 1033
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.603578154425612e-05,
+      "loss": 0.3525,
+      "step": 1034
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.601224105461394e-05,
+      "loss": 0.3797,
+      "step": 1035
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5988700564971752e-05,
+      "loss": 0.374,
+      "step": 1036
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.596516007532957e-05,
+      "loss": 0.3712,
+      "step": 1037
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5941619585687383e-05,
+      "loss": 0.3988,
+      "step": 1038
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.59180790960452e-05,
+      "loss": 0.3686,
+      "step": 1039
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.589453860640301e-05,
+      "loss": 0.3731,
+      "step": 1040
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.587099811676083e-05,
+      "loss": 0.3765,
+      "step": 1041
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5847457627118642e-05,
+      "loss": 0.3749,
+      "step": 1042
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.582391713747646e-05,
+      "loss": 0.3809,
+      "step": 1043
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5800376647834273e-05,
+      "loss": 0.3612,
+      "step": 1044
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5776836158192092e-05,
+      "loss": 0.3807,
+      "step": 1045
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5753295668549908e-05,
+      "loss": 0.3735,
+      "step": 1046
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5729755178907723e-05,
+      "loss": 0.3979,
+      "step": 1047
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.570621468926554e-05,
+      "loss": 0.3822,
+      "step": 1048
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.568267419962335e-05,
+      "loss": 0.3717,
+      "step": 1049
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.565913370998117e-05,
+      "loss": 0.3787,
+      "step": 1050
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5635593220338982e-05,
+      "loss": 0.3757,
+      "step": 1051
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.56120527306968e-05,
+      "loss": 0.3889,
+      "step": 1052
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5588512241054613e-05,
+      "loss": 0.367,
+      "step": 1053
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5564971751412432e-05,
+      "loss": 0.377,
+      "step": 1054
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5541431261770244e-05,
+      "loss": 0.3475,
+      "step": 1055
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5517890772128063e-05,
+      "loss": 0.3738,
+      "step": 1056
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.549435028248588e-05,
+      "loss": 0.3873,
+      "step": 1057
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.547080979284369e-05,
+      "loss": 0.3645,
+      "step": 1058
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.544726930320151e-05,
+      "loss": 0.3891,
+      "step": 1059
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5423728813559322e-05,
+      "loss": 0.3819,
+      "step": 1060
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.540018832391714e-05,
+      "loss": 0.3626,
+      "step": 1061
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5376647834274953e-05,
+      "loss": 0.3597,
+      "step": 1062
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.535310734463277e-05,
+      "loss": 0.3555,
+      "step": 1063
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5329566854990584e-05,
+      "loss": 0.37,
+      "step": 1064
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.53060263653484e-05,
+      "loss": 0.3865,
+      "step": 1065
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5282485875706215e-05,
+      "loss": 0.3754,
+      "step": 1066
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.525894538606403e-05,
+      "loss": 0.3664,
+      "step": 1067
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.523540489642185e-05,
+      "loss": 0.3828,
+      "step": 1068
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.521186440677966e-05,
+      "loss": 0.3708,
+      "step": 1069
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.518832391713748e-05,
+      "loss": 0.3684,
+      "step": 1070
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5164783427495293e-05,
+      "loss": 0.3746,
+      "step": 1071
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.514124293785311e-05,
+      "loss": 0.3828,
+      "step": 1072
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5117702448210924e-05,
+      "loss": 0.3772,
+      "step": 1073
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.509416195856874e-05,
+      "loss": 0.3688,
+      "step": 1074
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.507062146892655e-05,
+      "loss": 0.3842,
+      "step": 1075
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.504708097928437e-05,
+      "loss": 0.3883,
+      "step": 1076
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5023540489642182e-05,
+      "loss": 0.3609,
+      "step": 1077
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5e-05,
+      "loss": 0.3941,
+      "step": 1078
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4976459510357817e-05,
+      "loss": 0.383,
+      "step": 1079
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4952919020715632e-05,
+      "loss": 0.3601,
+      "step": 1080
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4929378531073448e-05,
+      "loss": 0.3881,
+      "step": 1081
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4905838041431263e-05,
+      "loss": 0.3599,
+      "step": 1082
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.488229755178908e-05,
+      "loss": 0.386,
+      "step": 1083
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4858757062146894e-05,
+      "loss": 0.3789,
+      "step": 1084
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.483521657250471e-05,
+      "loss": 0.3761,
+      "step": 1085
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4811676082862525e-05,
+      "loss": 0.3938,
+      "step": 1086
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.478813559322034e-05,
+      "loss": 0.3854,
+      "step": 1087
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4764595103578156e-05,
+      "loss": 0.3779,
+      "step": 1088
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4741054613935972e-05,
+      "loss": 0.3717,
+      "step": 1089
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4717514124293788e-05,
+      "loss": 0.3837,
+      "step": 1090
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4693973634651603e-05,
+      "loss": 0.3848,
+      "step": 1091
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4670433145009415e-05,
+      "loss": 0.3659,
+      "step": 1092
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.464689265536723e-05,
+      "loss": 0.3575,
+      "step": 1093
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4623352165725046e-05,
+      "loss": 0.3866,
+      "step": 1094
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4599811676082865e-05,
+      "loss": 0.3726,
+      "step": 1095
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.457627118644068e-05,
+      "loss": 0.3921,
+      "step": 1096
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4552730696798496e-05,
+      "loss": 0.3736,
+      "step": 1097
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4529190207156312e-05,
+      "loss": 0.4008,
+      "step": 1098
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4505649717514127e-05,
+      "loss": 0.3959,
+      "step": 1099
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.448210922787194e-05,
+      "loss": 0.3785,
+      "step": 1100
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4458568738229755e-05,
+      "loss": 0.3646,
+      "step": 1101
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.443502824858757e-05,
+      "loss": 0.3734,
+      "step": 1102
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4411487758945386e-05,
+      "loss": 0.3903,
+      "step": 1103
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.43879472693032e-05,
+      "loss": 0.3666,
+      "step": 1104
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4364406779661017e-05,
+      "loss": 0.3606,
+      "step": 1105
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4340866290018833e-05,
+      "loss": 0.3721,
+      "step": 1106
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.431732580037665e-05,
+      "loss": 0.3514,
+      "step": 1107
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4293785310734467e-05,
+      "loss": 0.396,
+      "step": 1108
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.427024482109228e-05,
+      "loss": 0.3624,
+      "step": 1109
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4246704331450095e-05,
+      "loss": 0.3767,
+      "step": 1110
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.422316384180791e-05,
+      "loss": 0.3516,
+      "step": 1111
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4199623352165726e-05,
+      "loss": 0.3846,
+      "step": 1112
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.417608286252354e-05,
+      "loss": 0.3347,
+      "step": 1113
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4152542372881357e-05,
+      "loss": 0.3757,
+      "step": 1114
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4129001883239172e-05,
+      "loss": 0.3927,
+      "step": 1115
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4105461393596988e-05,
+      "loss": 0.3794,
+      "step": 1116
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.4081920903954803e-05,
+      "loss": 0.3659,
+      "step": 1117
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.405838041431262e-05,
+      "loss": 0.3635,
+      "step": 1118
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.4034839924670434e-05,
+      "loss": 0.3717,
+      "step": 1119
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.401129943502825e-05,
+      "loss": 0.3741,
+      "step": 1120
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3987758945386065e-05,
+      "loss": 0.3878,
+      "step": 1121
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.396421845574388e-05,
+      "loss": 0.3774,
+      "step": 1122
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3940677966101697e-05,
+      "loss": 0.3602,
+      "step": 1123
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3917137476459512e-05,
+      "loss": 0.3658,
+      "step": 1124
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3893596986817328e-05,
+      "loss": 0.3646,
+      "step": 1125
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3870056497175143e-05,
+      "loss": 0.3444,
+      "step": 1126
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3846516007532955e-05,
+      "loss": 0.3778,
+      "step": 1127
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.382297551789077e-05,
+      "loss": 0.3695,
+      "step": 1128
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.379943502824859e-05,
+      "loss": 0.3831,
+      "step": 1129
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3775894538606405e-05,
+      "loss": 0.3757,
+      "step": 1130
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.375235404896422e-05,
+      "loss": 0.3811,
+      "step": 1131
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3728813559322036e-05,
+      "loss": 0.3621,
+      "step": 1132
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3705273069679852e-05,
+      "loss": 0.3757,
+      "step": 1133
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3681732580037667e-05,
+      "loss": 0.3892,
+      "step": 1134
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3658192090395483e-05,
+      "loss": 0.3824,
+      "step": 1135
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3634651600753295e-05,
+      "loss": 0.3774,
+      "step": 1136
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.361111111111111e-05,
+      "loss": 0.3694,
+      "step": 1137
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3587570621468926e-05,
+      "loss": 0.363,
+      "step": 1138
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.356403013182674e-05,
+      "loss": 0.3633,
+      "step": 1139
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3540489642184557e-05,
+      "loss": 0.3486,
+      "step": 1140
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3516949152542376e-05,
+      "loss": 0.3751,
+      "step": 1141
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.349340866290019e-05,
+      "loss": 0.3862,
+      "step": 1142
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3469868173258007e-05,
+      "loss": 0.3953,
+      "step": 1143
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.344632768361582e-05,
+      "loss": 0.3849,
+      "step": 1144
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3422787193973635e-05,
+      "loss": 0.3538,
+      "step": 1145
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.339924670433145e-05,
+      "loss": 0.3808,
+      "step": 1146
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3375706214689266e-05,
+      "loss": 0.401,
+      "step": 1147
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.335216572504708e-05,
+      "loss": 0.3698,
+      "step": 1148
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3328625235404897e-05,
+      "loss": 0.3592,
+      "step": 1149
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3305084745762712e-05,
+      "loss": 0.3759,
+      "step": 1150
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3281544256120528e-05,
+      "loss": 0.3671,
+      "step": 1151
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3258003766478343e-05,
+      "loss": 0.3602,
+      "step": 1152
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.323446327683616e-05,
+      "loss": 0.3655,
+      "step": 1153
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3210922787193975e-05,
+      "loss": 0.3732,
+      "step": 1154
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.318738229755179e-05,
+      "loss": 0.3579,
+      "step": 1155
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3163841807909606e-05,
+      "loss": 0.3712,
+      "step": 1156
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.314030131826742e-05,
+      "loss": 0.3682,
+      "step": 1157
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3116760828625237e-05,
+      "loss": 0.3754,
+      "step": 1158
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3093220338983052e-05,
+      "loss": 0.363,
+      "step": 1159
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.3069679849340868e-05,
+      "loss": 0.362,
+      "step": 1160
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.3046139359698683e-05,
+      "loss": 0.3849,
+      "step": 1161
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.30225988700565e-05,
+      "loss": 0.3891,
+      "step": 1162
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2999058380414314e-05,
+      "loss": 0.3605,
+      "step": 1163
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.297551789077213e-05,
+      "loss": 0.3465,
+      "step": 1164
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2951977401129945e-05,
+      "loss": 0.3836,
+      "step": 1165
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.292843691148776e-05,
+      "loss": 0.3718,
+      "step": 1166
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2904896421845576e-05,
+      "loss": 0.3875,
+      "step": 1167
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2881355932203392e-05,
+      "loss": 0.3661,
+      "step": 1168
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2857815442561207e-05,
+      "loss": 0.3646,
+      "step": 1169
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2834274952919023e-05,
+      "loss": 0.3862,
+      "step": 1170
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2810734463276835e-05,
+      "loss": 0.35,
+      "step": 1171
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.278719397363465e-05,
+      "loss": 0.3913,
+      "step": 1172
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2763653483992466e-05,
+      "loss": 0.3736,
+      "step": 1173
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.274011299435028e-05,
+      "loss": 0.3278,
+      "step": 1174
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.27165725047081e-05,
+      "loss": 0.3563,
+      "step": 1175
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2693032015065916e-05,
+      "loss": 0.371,
+      "step": 1176
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.266949152542373e-05,
+      "loss": 0.3551,
+      "step": 1177
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2645951035781547e-05,
+      "loss": 0.349,
+      "step": 1178
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.262241054613936e-05,
+      "loss": 0.3556,
+      "step": 1179
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2598870056497175e-05,
+      "loss": 0.36,
+      "step": 1180
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.257532956685499e-05,
+      "loss": 0.3792,
+      "step": 1181
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2551789077212806e-05,
+      "loss": 0.3714,
+      "step": 1182
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.252824858757062e-05,
+      "loss": 0.3569,
+      "step": 1183
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2504708097928437e-05,
+      "loss": 0.3678,
+      "step": 1184
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2481167608286252e-05,
+      "loss": 0.374,
+      "step": 1185
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.245762711864407e-05,
+      "loss": 0.3875,
+      "step": 1186
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2434086629001887e-05,
+      "loss": 0.3807,
+      "step": 1187
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.24105461393597e-05,
+      "loss": 0.3657,
+      "step": 1188
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2387005649717515e-05,
+      "loss": 0.3844,
+      "step": 1189
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.236346516007533e-05,
+      "loss": 0.3837,
+      "step": 1190
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2339924670433146e-05,
+      "loss": 0.3786,
+      "step": 1191
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.231638418079096e-05,
+      "loss": 0.3768,
+      "step": 1192
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2292843691148777e-05,
+      "loss": 0.3844,
+      "step": 1193
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2269303201506592e-05,
+      "loss": 0.3486,
+      "step": 1194
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2245762711864408e-05,
+      "loss": 0.3819,
+      "step": 1195
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2222222222222223e-05,
+      "loss": 0.3744,
+      "step": 1196
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.219868173258004e-05,
+      "loss": 0.3697,
+      "step": 1197
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2175141242937854e-05,
+      "loss": 0.3647,
+      "step": 1198
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.215160075329567e-05,
+      "loss": 0.3796,
+      "step": 1199
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2128060263653485e-05,
+      "loss": 0.3982,
+      "step": 1200
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.21045197740113e-05,
+      "loss": 0.4041,
+      "step": 1201
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2080979284369116e-05,
+      "loss": 0.3465,
+      "step": 1202
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.2057438794726932e-05,
+      "loss": 0.3917,
+      "step": 1203
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.2033898305084748e-05,
+      "loss": 0.362,
+      "step": 1204
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.2010357815442563e-05,
+      "loss": 0.3953,
+      "step": 1205
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1986817325800375e-05,
+      "loss": 0.3822,
+      "step": 1206
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.196327683615819e-05,
+      "loss": 0.3741,
+      "step": 1207
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.193973634651601e-05,
+      "loss": 0.4179,
+      "step": 1208
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1916195856873825e-05,
+      "loss": 0.3679,
+      "step": 1209
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.189265536723164e-05,
+      "loss": 0.3678,
+      "step": 1210
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1869114877589456e-05,
+      "loss": 0.3729,
+      "step": 1211
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1845574387947272e-05,
+      "loss": 0.3626,
+      "step": 1212
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1822033898305087e-05,
+      "loss": 0.3569,
+      "step": 1213
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1798493408662903e-05,
+      "loss": 0.3991,
+      "step": 1214
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1774952919020715e-05,
+      "loss": 0.3904,
+      "step": 1215
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.175141242937853e-05,
+      "loss": 0.3604,
+      "step": 1216
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1727871939736346e-05,
+      "loss": 0.3682,
+      "step": 1217
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.170433145009416e-05,
+      "loss": 0.3823,
+      "step": 1218
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1680790960451977e-05,
+      "loss": 0.3674,
+      "step": 1219
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1657250470809796e-05,
+      "loss": 0.375,
+      "step": 1220
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.163370998116761e-05,
+      "loss": 0.3701,
+      "step": 1221
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1610169491525427e-05,
+      "loss": 0.3653,
+      "step": 1222
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.158662900188324e-05,
+      "loss": 0.3634,
+      "step": 1223
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1563088512241055e-05,
+      "loss": 0.3617,
+      "step": 1224
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.153954802259887e-05,
+      "loss": 0.3716,
+      "step": 1225
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1516007532956686e-05,
+      "loss": 0.3803,
+      "step": 1226
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.14924670433145e-05,
+      "loss": 0.3657,
+      "step": 1227
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1468926553672317e-05,
+      "loss": 0.3728,
+      "step": 1228
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1445386064030132e-05,
+      "loss": 0.359,
+      "step": 1229
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1421845574387948e-05,
+      "loss": 0.3491,
+      "step": 1230
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1398305084745763e-05,
+      "loss": 0.3678,
+      "step": 1231
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.137476459510358e-05,
+      "loss": 0.3684,
+      "step": 1232
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1351224105461394e-05,
+      "loss": 0.3758,
+      "step": 1233
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.132768361581921e-05,
+      "loss": 0.3774,
+      "step": 1234
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1304143126177025e-05,
+      "loss": 0.3692,
+      "step": 1235
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.128060263653484e-05,
+      "loss": 0.3746,
+      "step": 1236
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1257062146892657e-05,
+      "loss": 0.3695,
+      "step": 1237
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1233521657250472e-05,
+      "loss": 0.3871,
+      "step": 1238
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1209981167608288e-05,
+      "loss": 0.3849,
+      "step": 1239
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1186440677966103e-05,
+      "loss": 0.3742,
+      "step": 1240
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.116290018832392e-05,
+      "loss": 0.3746,
+      "step": 1241
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1139359698681734e-05,
+      "loss": 0.3756,
+      "step": 1242
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.111581920903955e-05,
+      "loss": 0.369,
+      "step": 1243
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1092278719397365e-05,
+      "loss": 0.361,
+      "step": 1244
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.106873822975518e-05,
+      "loss": 0.3621,
+      "step": 1245
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.1045197740112996e-05,
+      "loss": 0.3718,
+      "step": 1246
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.1021657250470812e-05,
+      "loss": 0.359,
+      "step": 1247
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0998116760828627e-05,
+      "loss": 0.378,
+      "step": 1248
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0974576271186443e-05,
+      "loss": 0.3877,
+      "step": 1249
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0951035781544255e-05,
+      "loss": 0.3694,
+      "step": 1250
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.092749529190207e-05,
+      "loss": 0.3584,
+      "step": 1251
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0903954802259886e-05,
+      "loss": 0.4041,
+      "step": 1252
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.08804143126177e-05,
+      "loss": 0.3604,
+      "step": 1253
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.085687382297552e-05,
+      "loss": 0.3657,
+      "step": 1254
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0833333333333336e-05,
+      "loss": 0.3812,
+      "step": 1255
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.080979284369115e-05,
+      "loss": 0.3797,
+      "step": 1256
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0786252354048967e-05,
+      "loss": 0.3614,
+      "step": 1257
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.076271186440678e-05,
+      "loss": 0.3697,
+      "step": 1258
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0739171374764595e-05,
+      "loss": 0.3646,
+      "step": 1259
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.071563088512241e-05,
+      "loss": 0.3343,
+      "step": 1260
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0692090395480226e-05,
+      "loss": 0.3549,
+      "step": 1261
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.066854990583804e-05,
+      "loss": 0.37,
+      "step": 1262
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0645009416195857e-05,
+      "loss": 0.3861,
+      "step": 1263
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0621468926553672e-05,
+      "loss": 0.3581,
+      "step": 1264
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.059792843691149e-05,
+      "loss": 0.3599,
+      "step": 1265
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0574387947269307e-05,
+      "loss": 0.3826,
+      "step": 1266
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.055084745762712e-05,
+      "loss": 0.3773,
+      "step": 1267
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0527306967984934e-05,
+      "loss": 0.3669,
+      "step": 1268
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.050376647834275e-05,
+      "loss": 0.3587,
+      "step": 1269
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0480225988700566e-05,
+      "loss": 0.3934,
+      "step": 1270
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.045668549905838e-05,
+      "loss": 0.3674,
+      "step": 1271
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0433145009416197e-05,
+      "loss": 0.3894,
+      "step": 1272
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0409604519774012e-05,
+      "loss": 0.3849,
+      "step": 1273
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0386064030131828e-05,
+      "loss": 0.3702,
+      "step": 1274
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0362523540489643e-05,
+      "loss": 0.37,
+      "step": 1275
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.033898305084746e-05,
+      "loss": 0.3505,
+      "step": 1276
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0315442561205274e-05,
+      "loss": 0.3794,
+      "step": 1277
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.029190207156309e-05,
+      "loss": 0.3668,
+      "step": 1278
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0268361581920905e-05,
+      "loss": 0.368,
+      "step": 1279
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.024482109227872e-05,
+      "loss": 0.3841,
+      "step": 1280
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0221280602636536e-05,
+      "loss": 0.3668,
+      "step": 1281
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0197740112994352e-05,
+      "loss": 0.379,
+      "step": 1282
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0174199623352167e-05,
+      "loss": 0.3623,
+      "step": 1283
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0150659133709983e-05,
+      "loss": 0.3617,
+      "step": 1284
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0127118644067795e-05,
+      "loss": 0.3748,
+      "step": 1285
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.010357815442561e-05,
+      "loss": 0.3823,
+      "step": 1286
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0080037664783426e-05,
+      "loss": 0.3963,
+      "step": 1287
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0056497175141245e-05,
+      "loss": 0.356,
+      "step": 1288
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.003295668549906e-05,
+      "loss": 0.3457,
+      "step": 1289
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0009416195856876e-05,
+      "loss": 0.3744,
+      "step": 1290
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.998587570621469e-05,
+      "loss": 0.3897,
+      "step": 1291
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9962335216572507e-05,
+      "loss": 0.3889,
+      "step": 1292
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9938794726930323e-05,
+      "loss": 0.3892,
+      "step": 1293
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9915254237288135e-05,
+      "loss": 0.399,
+      "step": 1294
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.989171374764595e-05,
+      "loss": 0.378,
+      "step": 1295
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9868173258003766e-05,
+      "loss": 0.3615,
+      "step": 1296
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.984463276836158e-05,
+      "loss": 0.3768,
+      "step": 1297
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9821092278719397e-05,
+      "loss": 0.3681,
+      "step": 1298
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9797551789077216e-05,
+      "loss": 0.3587,
+      "step": 1299
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.977401129943503e-05,
+      "loss": 0.3659,
+      "step": 1300
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9750470809792847e-05,
+      "loss": 0.3686,
+      "step": 1301
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.972693032015066e-05,
+      "loss": 0.3677,
+      "step": 1302
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9703389830508475e-05,
+      "loss": 0.3707,
+      "step": 1303
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.967984934086629e-05,
+      "loss": 0.3647,
+      "step": 1304
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9656308851224106e-05,
+      "loss": 0.3738,
+      "step": 1305
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.963276836158192e-05,
+      "loss": 0.3781,
+      "step": 1306
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9609227871939737e-05,
+      "loss": 0.3539,
+      "step": 1307
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9585687382297552e-05,
+      "loss": 0.3651,
+      "step": 1308
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9562146892655368e-05,
+      "loss": 0.3808,
+      "step": 1309
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9538606403013183e-05,
+      "loss": 0.3559,
+      "step": 1310
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9515065913371e-05,
+      "loss": 0.367,
+      "step": 1311
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9491525423728814e-05,
+      "loss": 0.3846,
+      "step": 1312
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.946798493408663e-05,
+      "loss": 0.3564,
+      "step": 1313
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9444444444444445e-05,
+      "loss": 0.3714,
+      "step": 1314
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.942090395480226e-05,
+      "loss": 0.3607,
+      "step": 1315
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9397363465160076e-05,
+      "loss": 0.3674,
+      "step": 1316
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9373822975517892e-05,
+      "loss": 0.3689,
+      "step": 1317
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9350282485875707e-05,
+      "loss": 0.3609,
+      "step": 1318
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9326741996233523e-05,
+      "loss": 0.355,
+      "step": 1319
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.930320150659134e-05,
+      "loss": 0.3475,
+      "step": 1320
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9279661016949154e-05,
+      "loss": 0.3655,
+      "step": 1321
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.925612052730697e-05,
+      "loss": 0.3523,
+      "step": 1322
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9232580037664785e-05,
+      "loss": 0.3575,
+      "step": 1323
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.92090395480226e-05,
+      "loss": 0.359,
+      "step": 1324
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9185499058380416e-05,
+      "loss": 0.3682,
+      "step": 1325
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9161958568738232e-05,
+      "loss": 0.3923,
+      "step": 1326
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9138418079096047e-05,
+      "loss": 0.3362,
+      "step": 1327
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9114877589453863e-05,
+      "loss": 0.3602,
+      "step": 1328
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9091337099811675e-05,
+      "loss": 0.3719,
+      "step": 1329
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.906779661016949e-05,
+      "loss": 0.3643,
+      "step": 1330
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9044256120527306e-05,
+      "loss": 0.3653,
+      "step": 1331
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.902071563088512e-05,
+      "loss": 0.375,
+      "step": 1332
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.899717514124294e-05,
+      "loss": 0.3638,
+      "step": 1333
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8973634651600756e-05,
+      "loss": 0.3629,
+      "step": 1334
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.895009416195857e-05,
+      "loss": 0.3648,
+      "step": 1335
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8926553672316387e-05,
+      "loss": 0.3682,
+      "step": 1336
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.89030131826742e-05,
+      "loss": 0.3696,
+      "step": 1337
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8879472693032015e-05,
+      "loss": 0.3874,
+      "step": 1338
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.885593220338983e-05,
+      "loss": 0.3711,
+      "step": 1339
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8832391713747646e-05,
+      "loss": 0.4041,
+      "step": 1340
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.880885122410546e-05,
+      "loss": 0.3446,
+      "step": 1341
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8785310734463277e-05,
+      "loss": 0.3529,
+      "step": 1342
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8761770244821092e-05,
+      "loss": 0.38,
+      "step": 1343
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.873822975517891e-05,
+      "loss": 0.356,
+      "step": 1344
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8714689265536727e-05,
+      "loss": 0.3492,
+      "step": 1345
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.869114877589454e-05,
+      "loss": 0.3459,
+      "step": 1346
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8667608286252354e-05,
+      "loss": 0.3702,
+      "step": 1347
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.864406779661017e-05,
+      "loss": 0.3668,
+      "step": 1348
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8620527306967985e-05,
+      "loss": 0.365,
+      "step": 1349
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.85969868173258e-05,
+      "loss": 0.3696,
+      "step": 1350
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8573446327683617e-05,
+      "loss": 0.376,
+      "step": 1351
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8549905838041432e-05,
+      "loss": 0.3581,
+      "step": 1352
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8526365348399248e-05,
+      "loss": 0.399,
+      "step": 1353
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8502824858757063e-05,
+      "loss": 0.3742,
+      "step": 1354
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.847928436911488e-05,
+      "loss": 0.3908,
+      "step": 1355
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8455743879472694e-05,
+      "loss": 0.3622,
+      "step": 1356
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.843220338983051e-05,
+      "loss": 0.3788,
+      "step": 1357
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8408662900188325e-05,
+      "loss": 0.3618,
+      "step": 1358
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.838512241054614e-05,
+      "loss": 0.3535,
+      "step": 1359
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8361581920903956e-05,
+      "loss": 0.3671,
+      "step": 1360
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8338041431261772e-05,
+      "loss": 0.3593,
+      "step": 1361
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8314500941619587e-05,
+      "loss": 0.3792,
+      "step": 1362
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8290960451977403e-05,
+      "loss": 0.3563,
+      "step": 1363
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8267419962335215e-05,
+      "loss": 0.3485,
+      "step": 1364
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.824387947269303e-05,
+      "loss": 0.3915,
+      "step": 1365
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8220338983050846e-05,
+      "loss": 0.3985,
+      "step": 1366
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8196798493408665e-05,
+      "loss": 0.3336,
+      "step": 1367
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.817325800376648e-05,
+      "loss": 0.3544,
+      "step": 1368
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8149717514124296e-05,
+      "loss": 0.3759,
+      "step": 1369
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.812617702448211e-05,
+      "loss": 0.3687,
+      "step": 1370
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8102636534839927e-05,
+      "loss": 0.3782,
+      "step": 1371
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8079096045197743e-05,
+      "loss": 0.369,
+      "step": 1372
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8055555555555555e-05,
+      "loss": 0.3548,
+      "step": 1373
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.803201506591337e-05,
+      "loss": 0.356,
+      "step": 1374
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.8008474576271186e-05,
+      "loss": 0.3778,
+      "step": 1375
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7984934086629e-05,
+      "loss": 0.3628,
+      "step": 1376
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7961393596986817e-05,
+      "loss": 0.3686,
+      "step": 1377
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7937853107344636e-05,
+      "loss": 0.3507,
+      "step": 1378
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.791431261770245e-05,
+      "loss": 0.3638,
+      "step": 1379
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7890772128060267e-05,
+      "loss": 0.3488,
+      "step": 1380
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.786723163841808e-05,
+      "loss": 0.3706,
+      "step": 1381
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7843691148775894e-05,
+      "loss": 0.3743,
+      "step": 1382
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.782015065913371e-05,
+      "loss": 0.3699,
+      "step": 1383
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7796610169491526e-05,
+      "loss": 0.3685,
+      "step": 1384
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.777306967984934e-05,
+      "loss": 0.3743,
+      "step": 1385
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7749529190207157e-05,
+      "loss": 0.3566,
+      "step": 1386
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7725988700564972e-05,
+      "loss": 0.3635,
+      "step": 1387
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7702448210922788e-05,
+      "loss": 0.3593,
+      "step": 1388
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7678907721280603e-05,
+      "loss": 0.3634,
+      "step": 1389
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.765536723163842e-05,
+      "loss": 0.3459,
+      "step": 1390
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7631826741996234e-05,
+      "loss": 0.3828,
+      "step": 1391
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.760828625235405e-05,
+      "loss": 0.3381,
+      "step": 1392
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7584745762711865e-05,
+      "loss": 0.3586,
+      "step": 1393
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.756120527306968e-05,
+      "loss": 0.3768,
+      "step": 1394
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7537664783427496e-05,
+      "loss": 0.3706,
+      "step": 1395
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7514124293785312e-05,
+      "loss": 0.3444,
+      "step": 1396
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7490583804143127e-05,
+      "loss": 0.358,
+      "step": 1397
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7467043314500943e-05,
+      "loss": 0.3727,
+      "step": 1398
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.744350282485876e-05,
+      "loss": 0.3597,
+      "step": 1399
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7419962335216574e-05,
+      "loss": 0.3747,
+      "step": 1400
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.739642184557439e-05,
+      "loss": 0.3511,
+      "step": 1401
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7372881355932205e-05,
+      "loss": 0.3281,
+      "step": 1402
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.734934086629002e-05,
+      "loss": 0.377,
+      "step": 1403
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7325800376647836e-05,
+      "loss": 0.3748,
+      "step": 1404
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.730225988700565e-05,
+      "loss": 0.3737,
+      "step": 1405
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7278719397363467e-05,
+      "loss": 0.3515,
+      "step": 1406
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7255178907721283e-05,
+      "loss": 0.3617,
+      "step": 1407
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7231638418079095e-05,
+      "loss": 0.3621,
+      "step": 1408
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.720809792843691e-05,
+      "loss": 0.3702,
+      "step": 1409
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7184557438794726e-05,
+      "loss": 0.3725,
+      "step": 1410
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.716101694915254e-05,
+      "loss": 0.3688,
+      "step": 1411
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.713747645951036e-05,
+      "loss": 0.3559,
+      "step": 1412
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7113935969868176e-05,
+      "loss": 0.3497,
+      "step": 1413
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.709039548022599e-05,
+      "loss": 0.3639,
+      "step": 1414
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7066854990583807e-05,
+      "loss": 0.3815,
+      "step": 1415
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.704331450094162e-05,
+      "loss": 0.3635,
+      "step": 1416
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.7019774011299435e-05,
+      "loss": 0.369,
+      "step": 1417
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.699623352165725e-05,
+      "loss": 0.3515,
+      "step": 1418
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6972693032015066e-05,
+      "loss": 0.3618,
+      "step": 1419
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.694915254237288e-05,
+      "loss": 0.3681,
+      "step": 1420
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6925612052730697e-05,
+      "loss": 0.3856,
+      "step": 1421
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6902071563088512e-05,
+      "loss": 0.3384,
+      "step": 1422
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.687853107344633e-05,
+      "loss": 0.3599,
+      "step": 1423
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6854990583804147e-05,
+      "loss": 0.363,
+      "step": 1424
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.683145009416196e-05,
+      "loss": 0.3551,
+      "step": 1425
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6807909604519774e-05,
+      "loss": 0.3691,
+      "step": 1426
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.678436911487759e-05,
+      "loss": 0.3407,
+      "step": 1427
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6760828625235405e-05,
+      "loss": 0.2749,
+      "step": 1428
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.673728813559322e-05,
+      "loss": 0.2631,
+      "step": 1429
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6713747645951036e-05,
+      "loss": 0.2726,
+      "step": 1430
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6690207156308852e-05,
+      "loss": 0.2598,
+      "step": 1431
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.2445,
+      "step": 1432
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6643126177024483e-05,
+      "loss": 0.254,
+      "step": 1433
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.66195856873823e-05,
+      "loss": 0.2657,
+      "step": 1434
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6596045197740114e-05,
+      "loss": 0.2431,
+      "step": 1435
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.657250470809793e-05,
+      "loss": 0.2546,
+      "step": 1436
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6548964218455745e-05,
+      "loss": 0.2563,
+      "step": 1437
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.652542372881356e-05,
+      "loss": 0.2716,
+      "step": 1438
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6501883239171376e-05,
+      "loss": 0.2543,
+      "step": 1439
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.647834274952919e-05,
+      "loss": 0.2492,
+      "step": 1440
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6454802259887007e-05,
+      "loss": 0.248,
+      "step": 1441
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6431261770244823e-05,
+      "loss": 0.2522,
+      "step": 1442
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6407721280602635e-05,
+      "loss": 0.2394,
+      "step": 1443
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.638418079096045e-05,
+      "loss": 0.2254,
+      "step": 1444
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6360640301318266e-05,
+      "loss": 0.2433,
+      "step": 1445
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6337099811676085e-05,
+      "loss": 0.2483,
+      "step": 1446
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.63135593220339e-05,
+      "loss": 0.2425,
+      "step": 1447
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6290018832391716e-05,
+      "loss": 0.2273,
+      "step": 1448
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.626647834274953e-05,
+      "loss": 0.2414,
+      "step": 1449
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6242937853107347e-05,
+      "loss": 0.2385,
+      "step": 1450
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6219397363465162e-05,
+      "loss": 0.2473,
+      "step": 1451
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6195856873822975e-05,
+      "loss": 0.2416,
+      "step": 1452
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.617231638418079e-05,
+      "loss": 0.2645,
+      "step": 1453
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6148775894538606e-05,
+      "loss": 0.224,
+      "step": 1454
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.612523540489642e-05,
+      "loss": 0.2294,
+      "step": 1455
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6101694915254237e-05,
+      "loss": 0.2344,
+      "step": 1456
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6078154425612056e-05,
+      "loss": 0.241,
+      "step": 1457
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.605461393596987e-05,
+      "loss": 0.2428,
+      "step": 1458
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.6031073446327687e-05,
+      "loss": 0.2567,
+      "step": 1459
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.60075329566855e-05,
+      "loss": 0.269,
+      "step": 1460
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5983992467043314e-05,
+      "loss": 0.2386,
+      "step": 1461
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.596045197740113e-05,
+      "loss": 0.2416,
+      "step": 1462
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5936911487758945e-05,
+      "loss": 0.2561,
+      "step": 1463
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.591337099811676e-05,
+      "loss": 0.2519,
+      "step": 1464
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5889830508474576e-05,
+      "loss": 0.2344,
+      "step": 1465
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5866290018832392e-05,
+      "loss": 0.24,
+      "step": 1466
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5842749529190208e-05,
+      "loss": 0.2482,
+      "step": 1467
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5819209039548023e-05,
+      "loss": 0.2546,
+      "step": 1468
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.579566854990584e-05,
+      "loss": 0.2457,
+      "step": 1469
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5772128060263654e-05,
+      "loss": 0.231,
+      "step": 1470
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.574858757062147e-05,
+      "loss": 0.2417,
+      "step": 1471
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5725047080979285e-05,
+      "loss": 0.2534,
+      "step": 1472
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.57015065913371e-05,
+      "loss": 0.2426,
+      "step": 1473
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5677966101694916e-05,
+      "loss": 0.2381,
+      "step": 1474
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5654425612052732e-05,
+      "loss": 0.2528,
+      "step": 1475
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5630885122410547e-05,
+      "loss": 0.247,
+      "step": 1476
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5607344632768363e-05,
+      "loss": 0.2418,
+      "step": 1477
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5583804143126175e-05,
+      "loss": 0.25,
+      "step": 1478
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.556026365348399e-05,
+      "loss": 0.2571,
+      "step": 1479
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.553672316384181e-05,
+      "loss": 0.2363,
+      "step": 1480
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5513182674199625e-05,
+      "loss": 0.2291,
+      "step": 1481
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.548964218455744e-05,
+      "loss": 0.2314,
+      "step": 1482
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5466101694915256e-05,
+      "loss": 0.2231,
+      "step": 1483
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.544256120527307e-05,
+      "loss": 0.2318,
+      "step": 1484
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5419020715630887e-05,
+      "loss": 0.2408,
+      "step": 1485
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5395480225988703e-05,
+      "loss": 0.2255,
+      "step": 1486
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5371939736346515e-05,
+      "loss": 0.2452,
+      "step": 1487
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.534839924670433e-05,
+      "loss": 0.2351,
+      "step": 1488
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5324858757062146e-05,
+      "loss": 0.2567,
+      "step": 1489
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.530131826741996e-05,
+      "loss": 0.2474,
+      "step": 1490
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.527777777777778e-05,
+      "loss": 0.2332,
+      "step": 1491
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5254237288135596e-05,
+      "loss": 0.2435,
+      "step": 1492
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.523069679849341e-05,
+      "loss": 0.2309,
+      "step": 1493
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5207156308851225e-05,
+      "loss": 0.2632,
+      "step": 1494
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.518361581920904e-05,
+      "loss": 0.251,
+      "step": 1495
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5160075329566856e-05,
+      "loss": 0.2318,
+      "step": 1496
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5136534839924672e-05,
+      "loss": 0.2415,
+      "step": 1497
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5112994350282486e-05,
+      "loss": 0.2318,
+      "step": 1498
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5089453860640301e-05,
+      "loss": 0.2374,
+      "step": 1499
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5065913370998117e-05,
+      "loss": 0.2434,
+      "step": 1500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5042372881355932e-05,
+      "loss": 0.2395,
+      "step": 1501
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.501883239171375e-05,
+      "loss": 0.2376,
+      "step": 1502
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4995291902071565e-05,
+      "loss": 0.2356,
+      "step": 1503
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.497175141242938e-05,
+      "loss": 0.2477,
+      "step": 1504
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4948210922787196e-05,
+      "loss": 0.2442,
+      "step": 1505
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.492467043314501e-05,
+      "loss": 0.27,
+      "step": 1506
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4901129943502825e-05,
+      "loss": 0.2237,
+      "step": 1507
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.487758945386064e-05,
+      "loss": 0.2355,
+      "step": 1508
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4854048964218456e-05,
+      "loss": 0.2309,
+      "step": 1509
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4830508474576272e-05,
+      "loss": 0.2315,
+      "step": 1510
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4806967984934086e-05,
+      "loss": 0.2235,
+      "step": 1511
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4783427495291901e-05,
+      "loss": 0.2411,
+      "step": 1512
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.475988700564972e-05,
+      "loss": 0.235,
+      "step": 1513
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4736346516007536e-05,
+      "loss": 0.2406,
+      "step": 1514
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.471280602636535e-05,
+      "loss": 0.2409,
+      "step": 1515
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4689265536723165e-05,
+      "loss": 0.2459,
+      "step": 1516
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.466572504708098e-05,
+      "loss": 0.2245,
+      "step": 1517
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4642184557438796e-05,
+      "loss": 0.242,
+      "step": 1518
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4618644067796612e-05,
+      "loss": 0.2371,
+      "step": 1519
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4595103578154425e-05,
+      "loss": 0.2332,
+      "step": 1520
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4571563088512241e-05,
+      "loss": 0.2417,
+      "step": 1521
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4548022598870056e-05,
+      "loss": 0.2409,
+      "step": 1522
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4524482109227872e-05,
+      "loss": 0.2365,
+      "step": 1523
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4500941619585688e-05,
+      "loss": 0.2459,
+      "step": 1524
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4477401129943505e-05,
+      "loss": 0.2516,
+      "step": 1525
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.445386064030132e-05,
+      "loss": 0.2518,
+      "step": 1526
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4430320150659136e-05,
+      "loss": 0.241,
+      "step": 1527
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.440677966101695e-05,
+      "loss": 0.24,
+      "step": 1528
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4383239171374765e-05,
+      "loss": 0.2373,
+      "step": 1529
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.435969868173258e-05,
+      "loss": 0.2281,
+      "step": 1530
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4336158192090396e-05,
+      "loss": 0.2539,
+      "step": 1531
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4312617702448212e-05,
+      "loss": 0.2394,
+      "step": 1532
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4289077212806026e-05,
+      "loss": 0.2314,
+      "step": 1533
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4265536723163841e-05,
+      "loss": 0.2251,
+      "step": 1534
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4241996233521657e-05,
+      "loss": 0.2304,
+      "step": 1535
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4218455743879474e-05,
+      "loss": 0.2574,
+      "step": 1536
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.419491525423729e-05,
+      "loss": 0.2463,
+      "step": 1537
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4171374764595105e-05,
+      "loss": 0.2484,
+      "step": 1538
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.414783427495292e-05,
+      "loss": 0.236,
+      "step": 1539
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4124293785310736e-05,
+      "loss": 0.2364,
+      "step": 1540
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.410075329566855e-05,
+      "loss": 0.2236,
+      "step": 1541
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4077212806026365e-05,
+      "loss": 0.248,
+      "step": 1542
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4053672316384181e-05,
+      "loss": 0.2536,
+      "step": 1543
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4030131826741996e-05,
+      "loss": 0.2338,
+      "step": 1544
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.4006591337099812e-05,
+      "loss": 0.2378,
+      "step": 1545
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3983050847457627e-05,
+      "loss": 0.2421,
+      "step": 1546
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3959510357815445e-05,
+      "loss": 0.2457,
+      "step": 1547
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.393596986817326e-05,
+      "loss": 0.2318,
+      "step": 1548
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3912429378531076e-05,
+      "loss": 0.2338,
+      "step": 1549
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.388888888888889e-05,
+      "loss": 0.2404,
+      "step": 1550
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3865348399246705e-05,
+      "loss": 0.2498,
+      "step": 1551
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.384180790960452e-05,
+      "loss": 0.2616,
+      "step": 1552
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3818267419962336e-05,
+      "loss": 0.25,
+      "step": 1553
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3794726930320152e-05,
+      "loss": 0.2454,
+      "step": 1554
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3771186440677965e-05,
+      "loss": 0.2302,
+      "step": 1555
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3747645951035781e-05,
+      "loss": 0.2353,
+      "step": 1556
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3724105461393597e-05,
+      "loss": 0.2312,
+      "step": 1557
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3700564971751412e-05,
+      "loss": 0.2382,
+      "step": 1558
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.367702448210923e-05,
+      "loss": 0.2303,
+      "step": 1559
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3653483992467045e-05,
+      "loss": 0.2488,
+      "step": 1560
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.362994350282486e-05,
+      "loss": 0.2356,
+      "step": 1561
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3606403013182676e-05,
+      "loss": 0.2393,
+      "step": 1562
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.358286252354049e-05,
+      "loss": 0.2469,
+      "step": 1563
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3559322033898305e-05,
+      "loss": 0.2334,
+      "step": 1564
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.353578154425612e-05,
+      "loss": 0.235,
+      "step": 1565
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3512241054613936e-05,
+      "loss": 0.236,
+      "step": 1566
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3488700564971752e-05,
+      "loss": 0.2464,
+      "step": 1567
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3465160075329566e-05,
+      "loss": 0.256,
+      "step": 1568
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3441619585687381e-05,
+      "loss": 0.2407,
+      "step": 1569
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.34180790960452e-05,
+      "loss": 0.2411,
+      "step": 1570
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3394538606403016e-05,
+      "loss": 0.2357,
+      "step": 1571
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.337099811676083e-05,
+      "loss": 0.2378,
+      "step": 1572
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3347457627118645e-05,
+      "loss": 0.2422,
+      "step": 1573
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.332391713747646e-05,
+      "loss": 0.2491,
+      "step": 1574
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3300376647834276e-05,
+      "loss": 0.2161,
+      "step": 1575
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3276836158192092e-05,
+      "loss": 0.2399,
+      "step": 1576
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3253295668549905e-05,
+      "loss": 0.2337,
+      "step": 1577
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3229755178907721e-05,
+      "loss": 0.2344,
+      "step": 1578
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3206214689265536e-05,
+      "loss": 0.2428,
+      "step": 1579
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3182674199623352e-05,
+      "loss": 0.2517,
+      "step": 1580
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.315913370998117e-05,
+      "loss": 0.2267,
+      "step": 1581
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3135593220338985e-05,
+      "loss": 0.2351,
+      "step": 1582
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.31120527306968e-05,
+      "loss": 0.2497,
+      "step": 1583
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3088512241054616e-05,
+      "loss": 0.2409,
+      "step": 1584
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.306497175141243e-05,
+      "loss": 0.2401,
+      "step": 1585
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3041431261770245e-05,
+      "loss": 0.2374,
+      "step": 1586
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.301789077212806e-05,
+      "loss": 0.2269,
+      "step": 1587
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2994350282485876e-05,
+      "loss": 0.2502,
+      "step": 1588
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2970809792843692e-05,
+      "loss": 0.2324,
+      "step": 1589
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2947269303201506e-05,
+      "loss": 0.2482,
+      "step": 1590
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2923728813559321e-05,
+      "loss": 0.2586,
+      "step": 1591
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2900188323917137e-05,
+      "loss": 0.2457,
+      "step": 1592
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2876647834274954e-05,
+      "loss": 0.243,
+      "step": 1593
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.285310734463277e-05,
+      "loss": 0.2464,
+      "step": 1594
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2829566854990585e-05,
+      "loss": 0.2289,
+      "step": 1595
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.28060263653484e-05,
+      "loss": 0.2336,
+      "step": 1596
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2782485875706216e-05,
+      "loss": 0.2241,
+      "step": 1597
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2758945386064031e-05,
+      "loss": 0.2337,
+      "step": 1598
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2735404896421845e-05,
+      "loss": 0.2267,
+      "step": 1599
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2711864406779661e-05,
+      "loss": 0.2457,
+      "step": 1600
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2688323917137476e-05,
+      "loss": 0.2393,
+      "step": 1601
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2664783427495292e-05,
+      "loss": 0.2504,
+      "step": 1602
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2641242937853107e-05,
+      "loss": 0.2414,
+      "step": 1603
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2617702448210925e-05,
+      "loss": 0.2469,
+      "step": 1604
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.259416195856874e-05,
+      "loss": 0.2408,
+      "step": 1605
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2570621468926556e-05,
+      "loss": 0.2371,
+      "step": 1606
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.254708097928437e-05,
+      "loss": 0.2369,
+      "step": 1607
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2523540489642185e-05,
+      "loss": 0.239,
+      "step": 1608
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.25e-05,
+      "loss": 0.2507,
+      "step": 1609
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2476459510357816e-05,
+      "loss": 0.2471,
+      "step": 1610
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2452919020715632e-05,
+      "loss": 0.2407,
+      "step": 1611
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2429378531073447e-05,
+      "loss": 0.2443,
+      "step": 1612
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2405838041431263e-05,
+      "loss": 0.2267,
+      "step": 1613
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2382297551789078e-05,
+      "loss": 0.2485,
+      "step": 1614
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2358757062146894e-05,
+      "loss": 0.2425,
+      "step": 1615
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2335216572504708e-05,
+      "loss": 0.2276,
+      "step": 1616
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2311676082862523e-05,
+      "loss": 0.235,
+      "step": 1617
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.228813559322034e-05,
+      "loss": 0.219,
+      "step": 1618
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2264595103578156e-05,
+      "loss": 0.2365,
+      "step": 1619
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.224105461393597e-05,
+      "loss": 0.2451,
+      "step": 1620
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2217514124293785e-05,
+      "loss": 0.2317,
+      "step": 1621
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.21939736346516e-05,
+      "loss": 0.2286,
+      "step": 1622
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2170433145009416e-05,
+      "loss": 0.2416,
+      "step": 1623
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2146892655367234e-05,
+      "loss": 0.2468,
+      "step": 1624
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2123352165725047e-05,
+      "loss": 0.2361,
+      "step": 1625
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2099811676082863e-05,
+      "loss": 0.2386,
+      "step": 1626
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2076271186440678e-05,
+      "loss": 0.2371,
+      "step": 1627
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2052730696798494e-05,
+      "loss": 0.2463,
+      "step": 1628
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.202919020715631e-05,
+      "loss": 0.2308,
+      "step": 1629
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2005649717514125e-05,
+      "loss": 0.2439,
+      "step": 1630
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.198210922787194e-05,
+      "loss": 0.2371,
+      "step": 1631
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1958568738229756e-05,
+      "loss": 0.232,
+      "step": 1632
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1935028248587572e-05,
+      "loss": 0.2501,
+      "step": 1633
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1911487758945385e-05,
+      "loss": 0.2378,
+      "step": 1634
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1887947269303203e-05,
+      "loss": 0.2361,
+      "step": 1635
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1864406779661018e-05,
+      "loss": 0.234,
+      "step": 1636
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1840866290018834e-05,
+      "loss": 0.2345,
+      "step": 1637
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1817325800376648e-05,
+      "loss": 0.2475,
+      "step": 1638
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1793785310734463e-05,
+      "loss": 0.2336,
+      "step": 1639
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1770244821092279e-05,
+      "loss": 0.2317,
+      "step": 1640
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1746704331450096e-05,
+      "loss": 0.2381,
+      "step": 1641
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.172316384180791e-05,
+      "loss": 0.2326,
+      "step": 1642
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1699623352165725e-05,
+      "loss": 0.2349,
+      "step": 1643
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.167608286252354e-05,
+      "loss": 0.2238,
+      "step": 1644
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1652542372881356e-05,
+      "loss": 0.2466,
+      "step": 1645
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1629001883239172e-05,
+      "loss": 0.2533,
+      "step": 1646
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1605461393596987e-05,
+      "loss": 0.2292,
+      "step": 1647
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1581920903954803e-05,
+      "loss": 0.2584,
+      "step": 1648
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1558380414312618e-05,
+      "loss": 0.2357,
+      "step": 1649
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1534839924670434e-05,
+      "loss": 0.2496,
+      "step": 1650
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.151129943502825e-05,
+      "loss": 0.2428,
+      "step": 1651
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1487758945386065e-05,
+      "loss": 0.226,
+      "step": 1652
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.146421845574388e-05,
+      "loss": 0.2466,
+      "step": 1653
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1440677966101696e-05,
+      "loss": 0.2405,
+      "step": 1654
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1417137476459511e-05,
+      "loss": 0.2259,
+      "step": 1655
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1393596986817325e-05,
+      "loss": 0.2265,
+      "step": 1656
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.137005649717514e-05,
+      "loss": 0.2443,
+      "step": 1657
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1346516007532958e-05,
+      "loss": 0.2309,
+      "step": 1658
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1322975517890774e-05,
+      "loss": 0.2388,
+      "step": 1659
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1299435028248587e-05,
+      "loss": 0.2391,
+      "step": 1660
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1275894538606403e-05,
+      "loss": 0.2345,
+      "step": 1661
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1252354048964218e-05,
+      "loss": 0.2277,
+      "step": 1662
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1228813559322036e-05,
+      "loss": 0.245,
+      "step": 1663
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.120527306967985e-05,
+      "loss": 0.2534,
+      "step": 1664
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1181732580037665e-05,
+      "loss": 0.2213,
+      "step": 1665
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.115819209039548e-05,
+      "loss": 0.2283,
+      "step": 1666
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1134651600753296e-05,
+      "loss": 0.2443,
+      "step": 1667
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1111111111111112e-05,
+      "loss": 0.2406,
+      "step": 1668
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1087570621468927e-05,
+      "loss": 0.2226,
+      "step": 1669
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1064030131826743e-05,
+      "loss": 0.2369,
+      "step": 1670
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1040489642184558e-05,
+      "loss": 0.2258,
+      "step": 1671
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1016949152542374e-05,
+      "loss": 0.2334,
+      "step": 1672
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0993408662900188e-05,
+      "loss": 0.2274,
+      "step": 1673
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0969868173258005e-05,
+      "loss": 0.2398,
+      "step": 1674
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.094632768361582e-05,
+      "loss": 0.2326,
+      "step": 1675
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0922787193973636e-05,
+      "loss": 0.2329,
+      "step": 1676
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0899246704331451e-05,
+      "loss": 0.2252,
+      "step": 1677
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0875706214689265e-05,
+      "loss": 0.2308,
+      "step": 1678
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.085216572504708e-05,
+      "loss": 0.2413,
+      "step": 1679
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0828625235404898e-05,
+      "loss": 0.2346,
+      "step": 1680
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0805084745762714e-05,
+      "loss": 0.2292,
+      "step": 1681
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0781544256120527e-05,
+      "loss": 0.2513,
+      "step": 1682
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0758003766478343e-05,
+      "loss": 0.237,
+      "step": 1683
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0734463276836158e-05,
+      "loss": 0.2293,
+      "step": 1684
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0710922787193974e-05,
+      "loss": 0.2349,
+      "step": 1685
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.068738229755179e-05,
+      "loss": 0.2421,
+      "step": 1686
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0663841807909605e-05,
+      "loss": 0.2481,
+      "step": 1687
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.064030131826742e-05,
+      "loss": 0.2447,
+      "step": 1688
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0616760828625236e-05,
+      "loss": 0.2439,
+      "step": 1689
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0593220338983052e-05,
+      "loss": 0.232,
+      "step": 1690
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0569679849340867e-05,
+      "loss": 0.2309,
+      "step": 1691
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0546139359698683e-05,
+      "loss": 0.2396,
+      "step": 1692
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0522598870056498e-05,
+      "loss": 0.2483,
+      "step": 1693
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0499058380414314e-05,
+      "loss": 0.249,
+      "step": 1694
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0475517890772127e-05,
+      "loss": 0.2298,
+      "step": 1695
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0451977401129943e-05,
+      "loss": 0.2283,
+      "step": 1696
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.042843691148776e-05,
+      "loss": 0.2256,
+      "step": 1697
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0404896421845576e-05,
+      "loss": 0.229,
+      "step": 1698
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.038135593220339e-05,
+      "loss": 0.2601,
+      "step": 1699
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0357815442561205e-05,
+      "loss": 0.2346,
+      "step": 1700
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.033427495291902e-05,
+      "loss": 0.2263,
+      "step": 1701
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0310734463276836e-05,
+      "loss": 0.2356,
+      "step": 1702
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0287193973634653e-05,
+      "loss": 0.2355,
+      "step": 1703
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0263653483992467e-05,
+      "loss": 0.2428,
+      "step": 1704
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0240112994350283e-05,
+      "loss": 0.2357,
+      "step": 1705
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0216572504708098e-05,
+      "loss": 0.2508,
+      "step": 1706
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0193032015065914e-05,
+      "loss": 0.2737,
+      "step": 1707
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.016949152542373e-05,
+      "loss": 0.2505,
+      "step": 1708
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0145951035781545e-05,
+      "loss": 0.2556,
+      "step": 1709
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.012241054613936e-05,
+      "loss": 0.235,
+      "step": 1710
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0098870056497176e-05,
+      "loss": 0.2489,
+      "step": 1711
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0075329566854991e-05,
+      "loss": 0.2339,
+      "step": 1712
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0051789077212805e-05,
+      "loss": 0.2385,
+      "step": 1713
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0028248587570623e-05,
+      "loss": 0.2366,
+      "step": 1714
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0004708097928438e-05,
+      "loss": 0.2481,
+      "step": 1715
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.981167608286254e-06,
+      "loss": 0.2341,
+      "step": 1716
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.957627118644067e-06,
+      "loss": 0.2198,
+      "step": 1717
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.934086629001883e-06,
+      "loss": 0.2473,
+      "step": 1718
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.910546139359698e-06,
+      "loss": 0.2513,
+      "step": 1719
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.887005649717516e-06,
+      "loss": 0.2334,
+      "step": 1720
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.86346516007533e-06,
+      "loss": 0.2406,
+      "step": 1721
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.839924670433145e-06,
+      "loss": 0.2383,
+      "step": 1722
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.81638418079096e-06,
+      "loss": 0.2327,
+      "step": 1723
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.792843691148776e-06,
+      "loss": 0.2446,
+      "step": 1724
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.769303201506592e-06,
+      "loss": 0.2343,
+      "step": 1725
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.745762711864407e-06,
+      "loss": 0.2475,
+      "step": 1726
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.722222222222223e-06,
+      "loss": 0.2453,
+      "step": 1727
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.698681732580038e-06,
+      "loss": 0.2493,
+      "step": 1728
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.675141242937854e-06,
+      "loss": 0.2345,
+      "step": 1729
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.65160075329567e-06,
+      "loss": 0.2504,
+      "step": 1730
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.628060263653485e-06,
+      "loss": 0.2262,
+      "step": 1731
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.6045197740113e-06,
+      "loss": 0.228,
+      "step": 1732
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.580979284369116e-06,
+      "loss": 0.2396,
+      "step": 1733
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.557438794726931e-06,
+      "loss": 0.2289,
+      "step": 1734
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.533898305084745e-06,
+      "loss": 0.2373,
+      "step": 1735
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.51035781544256e-06,
+      "loss": 0.2244,
+      "step": 1736
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.486817325800378e-06,
+      "loss": 0.2264,
+      "step": 1737
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.463276836158193e-06,
+      "loss": 0.2358,
+      "step": 1738
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.439736346516007e-06,
+      "loss": 0.2459,
+      "step": 1739
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.416195856873823e-06,
+      "loss": 0.2219,
+      "step": 1740
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.392655367231638e-06,
+      "loss": 0.2476,
+      "step": 1741
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.369114877589456e-06,
+      "loss": 0.2563,
+      "step": 1742
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.34557438794727e-06,
+      "loss": 0.2469,
+      "step": 1743
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.322033898305085e-06,
+      "loss": 0.2228,
+      "step": 1744
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.2984934086629e-06,
+      "loss": 0.242,
+      "step": 1745
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.274952919020716e-06,
+      "loss": 0.2364,
+      "step": 1746
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.251412429378532e-06,
+      "loss": 0.2242,
+      "step": 1747
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.227871939736347e-06,
+      "loss": 0.24,
+      "step": 1748
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.204331450094163e-06,
+      "loss": 0.2417,
+      "step": 1749
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.180790960451978e-06,
+      "loss": 0.25,
+      "step": 1750
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.157250470809794e-06,
+      "loss": 0.2273,
+      "step": 1751
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.133709981167607e-06,
+      "loss": 0.2224,
+      "step": 1752
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.110169491525423e-06,
+      "loss": 0.2357,
+      "step": 1753
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.08662900188324e-06,
+      "loss": 0.2226,
+      "step": 1754
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.063088512241056e-06,
+      "loss": 0.2473,
+      "step": 1755
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.039548022598871e-06,
+      "loss": 0.2272,
+      "step": 1756
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.016007532956685e-06,
+      "loss": 0.2529,
+      "step": 1757
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.9924670433145e-06,
+      "loss": 0.2387,
+      "step": 1758
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.968926553672318e-06,
+      "loss": 0.2322,
+      "step": 1759
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.945386064030133e-06,
+      "loss": 0.2415,
+      "step": 1760
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.921845574387947e-06,
+      "loss": 0.2325,
+      "step": 1761
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.898305084745763e-06,
+      "loss": 0.225,
+      "step": 1762
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.874764595103578e-06,
+      "loss": 0.2314,
+      "step": 1763
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.851224105461394e-06,
+      "loss": 0.2158,
+      "step": 1764
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.82768361581921e-06,
+      "loss": 0.2432,
+      "step": 1765
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.804143126177025e-06,
+      "loss": 0.2352,
+      "step": 1766
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.78060263653484e-06,
+      "loss": 0.2321,
+      "step": 1767
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.757062146892656e-06,
+      "loss": 0.2212,
+      "step": 1768
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.733521657250471e-06,
+      "loss": 0.2223,
+      "step": 1769
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.709981167608287e-06,
+      "loss": 0.2312,
+      "step": 1770
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.686440677966103e-06,
+      "loss": 0.239,
+      "step": 1771
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.662900188323918e-06,
+      "loss": 0.253,
+      "step": 1772
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.639359698681734e-06,
+      "loss": 0.2383,
+      "step": 1773
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.615819209039547e-06,
+      "loss": 0.222,
+      "step": 1774
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.592278719397363e-06,
+      "loss": 0.24,
+      "step": 1775
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.56873822975518e-06,
+      "loss": 0.2399,
+      "step": 1776
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.545197740112996e-06,
+      "loss": 0.2434,
+      "step": 1777
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.52165725047081e-06,
+      "loss": 0.2343,
+      "step": 1778
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.498116760828625e-06,
+      "loss": 0.2285,
+      "step": 1779
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.47457627118644e-06,
+      "loss": 0.2501,
+      "step": 1780
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.451035781544256e-06,
+      "loss": 0.2362,
+      "step": 1781
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.427495291902073e-06,
+      "loss": 0.2245,
+      "step": 1782
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.403954802259887e-06,
+      "loss": 0.2347,
+      "step": 1783
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.380414312617703e-06,
+      "loss": 0.2128,
+      "step": 1784
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.356873822975518e-06,
+      "loss": 0.2411,
+      "step": 1785
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.2294,
+      "step": 1786
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.30979284369115e-06,
+      "loss": 0.2315,
+      "step": 1787
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.286252354048965e-06,
+      "loss": 0.2537,
+      "step": 1788
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.26271186440678e-06,
+      "loss": 0.2282,
+      "step": 1789
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.239171374764596e-06,
+      "loss": 0.2366,
+      "step": 1790
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.215630885122411e-06,
+      "loss": 0.2143,
+      "step": 1791
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.192090395480225e-06,
+      "loss": 0.2447,
+      "step": 1792
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.168549905838042e-06,
+      "loss": 0.2325,
+      "step": 1793
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.145009416195858e-06,
+      "loss": 0.2302,
+      "step": 1794
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.121468926553673e-06,
+      "loss": 0.2315,
+      "step": 1795
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.097928436911487e-06,
+      "loss": 0.2412,
+      "step": 1796
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.074387947269303e-06,
+      "loss": 0.2364,
+      "step": 1797
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.050847457627118e-06,
+      "loss": 0.2327,
+      "step": 1798
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.027306967984936e-06,
+      "loss": 0.2318,
+      "step": 1799
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.00376647834275e-06,
+      "loss": 0.2327,
+      "step": 1800
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.980225988700565e-06,
+      "loss": 0.2403,
+      "step": 1801
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.95668549905838e-06,
+      "loss": 0.2327,
+      "step": 1802
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.933145009416196e-06,
+      "loss": 0.2277,
+      "step": 1803
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.909604519774012e-06,
+      "loss": 0.2428,
+      "step": 1804
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.886064030131827e-06,
+      "loss": 0.2391,
+      "step": 1805
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.862523540489643e-06,
+      "loss": 0.2276,
+      "step": 1806
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.838983050847458e-06,
+      "loss": 0.2277,
+      "step": 1807
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.815442561205274e-06,
+      "loss": 0.2348,
+      "step": 1808
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.791902071563087e-06,
+      "loss": 0.2194,
+      "step": 1809
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.768361581920905e-06,
+      "loss": 0.2353,
+      "step": 1810
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.74482109227872e-06,
+      "loss": 0.2325,
+      "step": 1811
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.721280602636536e-06,
+      "loss": 0.2259,
+      "step": 1812
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.697740112994351e-06,
+      "loss": 0.2469,
+      "step": 1813
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.674199623352165e-06,
+      "loss": 0.2304,
+      "step": 1814
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.65065913370998e-06,
+      "loss": 0.2266,
+      "step": 1815
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.627118644067798e-06,
+      "loss": 0.2414,
+      "step": 1816
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.6035781544256125e-06,
+      "loss": 0.2418,
+      "step": 1817
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.580037664783428e-06,
+      "loss": 0.2244,
+      "step": 1818
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.556497175141243e-06,
+      "loss": 0.2117,
+      "step": 1819
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.532956685499058e-06,
+      "loss": 0.2411,
+      "step": 1820
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.509416195856875e-06,
+      "loss": 0.2187,
+      "step": 1821
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.48587570621469e-06,
+      "loss": 0.2441,
+      "step": 1822
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.462335216572505e-06,
+      "loss": 0.2379,
+      "step": 1823
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.43879472693032e-06,
+      "loss": 0.2382,
+      "step": 1824
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.415254237288136e-06,
+      "loss": 0.2464,
+      "step": 1825
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.391713747645951e-06,
+      "loss": 0.2352,
+      "step": 1826
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.368173258003768e-06,
+      "loss": 0.2356,
+      "step": 1827
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.3446327683615825e-06,
+      "loss": 0.2322,
+      "step": 1828
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.321092278719398e-06,
+      "loss": 0.2335,
+      "step": 1829
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.297551789077213e-06,
+      "loss": 0.2534,
+      "step": 1830
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.274011299435028e-06,
+      "loss": 0.2381,
+      "step": 1831
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.250470809792844e-06,
+      "loss": 0.2218,
+      "step": 1832
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.22693032015066e-06,
+      "loss": 0.2453,
+      "step": 1833
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.203389830508475e-06,
+      "loss": 0.2244,
+      "step": 1834
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.17984934086629e-06,
+      "loss": 0.2364,
+      "step": 1835
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.156308851224106e-06,
+      "loss": 0.2339,
+      "step": 1836
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.1327683615819206e-06,
+      "loss": 0.2446,
+      "step": 1837
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.109227871939737e-06,
+      "loss": 0.2415,
+      "step": 1838
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.0856873822975525e-06,
+      "loss": 0.2314,
+      "step": 1839
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.062146892655368e-06,
+      "loss": 0.2254,
+      "step": 1840
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.038606403013183e-06,
+      "loss": 0.236,
+      "step": 1841
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.015065913370998e-06,
+      "loss": 0.2211,
+      "step": 1842
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.991525423728814e-06,
+      "loss": 0.2403,
+      "step": 1843
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.96798493408663e-06,
+      "loss": 0.2459,
+      "step": 1844
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.944444444444445e-06,
+      "loss": 0.2251,
+      "step": 1845
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.92090395480226e-06,
+      "loss": 0.2344,
+      "step": 1846
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.897363465160076e-06,
+      "loss": 0.2314,
+      "step": 1847
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.8738229755178905e-06,
+      "loss": 0.2361,
+      "step": 1848
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.850282485875706e-06,
+      "loss": 0.2402,
+      "step": 1849
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.826741996233522e-06,
+      "loss": 0.244,
+      "step": 1850
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.803201506591338e-06,
+      "loss": 0.217,
+      "step": 1851
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.779661016949153e-06,
+      "loss": 0.2243,
+      "step": 1852
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.756120527306968e-06,
+      "loss": 0.2325,
+      "step": 1853
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.732580037664783e-06,
+      "loss": 0.2316,
+      "step": 1854
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.7090395480226e-06,
+      "loss": 0.2229,
+      "step": 1855
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.685499058380415e-06,
+      "loss": 0.2307,
+      "step": 1856
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.66195856873823e-06,
+      "loss": 0.2313,
+      "step": 1857
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.638418079096046e-06,
+      "loss": 0.227,
+      "step": 1858
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.6148775894538605e-06,
+      "loss": 0.2324,
+      "step": 1859
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.591337099811676e-06,
+      "loss": 0.2309,
+      "step": 1860
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.567796610169492e-06,
+      "loss": 0.2345,
+      "step": 1861
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.544256120527308e-06,
+      "loss": 0.2536,
+      "step": 1862
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.520715630885123e-06,
+      "loss": 0.2475,
+      "step": 1863
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.497175141242938e-06,
+      "loss": 0.2404,
+      "step": 1864
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.473634651600753e-06,
+      "loss": 0.2421,
+      "step": 1865
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.450094161958568e-06,
+      "loss": 0.2289,
+      "step": 1866
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.426553672316385e-06,
+      "loss": 0.2292,
+      "step": 1867
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.4030131826742e-06,
+      "loss": 0.247,
+      "step": 1868
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.379472693032016e-06,
+      "loss": 0.2264,
+      "step": 1869
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.3559322033898304e-06,
+      "loss": 0.2278,
+      "step": 1870
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.332391713747646e-06,
+      "loss": 0.2378,
+      "step": 1871
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.308851224105462e-06,
+      "loss": 0.2243,
+      "step": 1872
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.285310734463278e-06,
+      "loss": 0.2447,
+      "step": 1873
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.2617702448210925e-06,
+      "loss": 0.2313,
+      "step": 1874
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.238229755178908e-06,
+      "loss": 0.2171,
+      "step": 1875
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.214689265536724e-06,
+      "loss": 0.242,
+      "step": 1876
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.191148775894539e-06,
+      "loss": 0.2395,
+      "step": 1877
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.167608286252354e-06,
+      "loss": 0.2265,
+      "step": 1878
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.14406779661017e-06,
+      "loss": 0.2263,
+      "step": 1879
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.120527306967985e-06,
+      "loss": 0.2234,
+      "step": 1880
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.0969868173258e-06,
+      "loss": 0.2268,
+      "step": 1881
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.073446327683617e-06,
+      "loss": 0.2192,
+      "step": 1882
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.0499058380414314e-06,
+      "loss": 0.2155,
+      "step": 1883
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.026365348399247e-06,
+      "loss": 0.2353,
+      "step": 1884
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.0028248587570625e-06,
+      "loss": 0.2439,
+      "step": 1885
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.979284369114878e-06,
+      "loss": 0.2234,
+      "step": 1886
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.955743879472693e-06,
+      "loss": 0.2419,
+      "step": 1887
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.932203389830509e-06,
+      "loss": 0.2398,
+      "step": 1888
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.908662900188324e-06,
+      "loss": 0.2237,
+      "step": 1889
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.885122410546139e-06,
+      "loss": 0.2413,
+      "step": 1890
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.861581920903955e-06,
+      "loss": 0.2248,
+      "step": 1891
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.83804143126177e-06,
+      "loss": 0.2134,
+      "step": 1892
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.814500941619586e-06,
+      "loss": 0.2616,
+      "step": 1893
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.790960451977401e-06,
+      "loss": 0.2204,
+      "step": 1894
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.767419962335217e-06,
+      "loss": 0.2437,
+      "step": 1895
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.7438794726930324e-06,
+      "loss": 0.2305,
+      "step": 1896
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.720338983050848e-06,
+      "loss": 0.2351,
+      "step": 1897
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.696798493408663e-06,
+      "loss": 0.2209,
+      "step": 1898
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.673258003766479e-06,
+      "loss": 0.2378,
+      "step": 1899
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.649717514124294e-06,
+      "loss": 0.225,
+      "step": 1900
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.626177024482109e-06,
+      "loss": 0.2259,
+      "step": 1901
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.602636534839925e-06,
+      "loss": 0.2382,
+      "step": 1902
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.57909604519774e-06,
+      "loss": 0.2229,
+      "step": 1903
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.555555555555556e-06,
+      "loss": 0.2379,
+      "step": 1904
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.532015065913371e-06,
+      "loss": 0.2245,
+      "step": 1905
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.508474576271187e-06,
+      "loss": 0.2312,
+      "step": 1906
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.484934086629002e-06,
+      "loss": 0.2074,
+      "step": 1907
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.461393596986818e-06,
+      "loss": 0.232,
+      "step": 1908
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.437853107344633e-06,
+      "loss": 0.2389,
+      "step": 1909
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.414312617702449e-06,
+      "loss": 0.221,
+      "step": 1910
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.390772128060264e-06,
+      "loss": 0.2267,
+      "step": 1911
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.367231638418079e-06,
+      "loss": 0.2395,
+      "step": 1912
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.343691148775895e-06,
+      "loss": 0.2395,
+      "step": 1913
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.32015065913371e-06,
+      "loss": 0.2302,
+      "step": 1914
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.296610169491526e-06,
+      "loss": 0.2189,
+      "step": 1915
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.273069679849341e-06,
+      "loss": 0.2129,
+      "step": 1916
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.249529190207157e-06,
+      "loss": 0.2144,
+      "step": 1917
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.2259887005649715e-06,
+      "loss": 0.2329,
+      "step": 1918
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.202448210922788e-06,
+      "loss": 0.2468,
+      "step": 1919
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.1789077212806026e-06,
+      "loss": 0.2345,
+      "step": 1920
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.155367231638418e-06,
+      "loss": 0.2199,
+      "step": 1921
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.131826741996234e-06,
+      "loss": 0.2328,
+      "step": 1922
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.108286252354049e-06,
+      "loss": 0.2455,
+      "step": 1923
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.084745762711865e-06,
+      "loss": 0.2553,
+      "step": 1924
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.06120527306968e-06,
+      "loss": 0.2283,
+      "step": 1925
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.037664783427496e-06,
+      "loss": 0.2263,
+      "step": 1926
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.014124293785311e-06,
+      "loss": 0.2212,
+      "step": 1927
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.990583804143127e-06,
+      "loss": 0.2296,
+      "step": 1928
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.9670433145009415e-06,
+      "loss": 0.2241,
+      "step": 1929
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.943502824858758e-06,
+      "loss": 0.2174,
+      "step": 1930
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.9199623352165725e-06,
+      "loss": 0.242,
+      "step": 1931
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.896421845574388e-06,
+      "loss": 0.2273,
+      "step": 1932
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.872881355932204e-06,
+      "loss": 0.2331,
+      "step": 1933
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.849340866290019e-06,
+      "loss": 0.2319,
+      "step": 1934
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.825800376647835e-06,
+      "loss": 0.2194,
+      "step": 1935
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.80225988700565e-06,
+      "loss": 0.2402,
+      "step": 1936
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.778719397363466e-06,
+      "loss": 0.2089,
+      "step": 1937
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.75517890772128e-06,
+      "loss": 0.2224,
+      "step": 1938
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.731638418079097e-06,
+      "loss": 0.2243,
+      "step": 1939
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.7080979284369114e-06,
+      "loss": 0.2273,
+      "step": 1940
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.684557438794728e-06,
+      "loss": 0.2332,
+      "step": 1941
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6610169491525425e-06,
+      "loss": 0.232,
+      "step": 1942
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.637476459510358e-06,
+      "loss": 0.2582,
+      "step": 1943
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6139359698681735e-06,
+      "loss": 0.2366,
+      "step": 1944
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.590395480225989e-06,
+      "loss": 0.211,
+      "step": 1945
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.566854990583804e-06,
+      "loss": 0.2349,
+      "step": 1946
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.54331450094162e-06,
+      "loss": 0.2151,
+      "step": 1947
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.519774011299436e-06,
+      "loss": 0.2147,
+      "step": 1948
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.49623352165725e-06,
+      "loss": 0.2438,
+      "step": 1949
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.472693032015067e-06,
+      "loss": 0.2327,
+      "step": 1950
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.449152542372881e-06,
+      "loss": 0.2389,
+      "step": 1951
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.425612052730697e-06,
+      "loss": 0.2377,
+      "step": 1952
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.4020715630885124e-06,
+      "loss": 0.2458,
+      "step": 1953
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.378531073446328e-06,
+      "loss": 0.2376,
+      "step": 1954
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3549905838041435e-06,
+      "loss": 0.2361,
+      "step": 1955
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.331450094161959e-06,
+      "loss": 0.2349,
+      "step": 1956
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.307909604519774e-06,
+      "loss": 0.2269,
+      "step": 1957
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.28436911487759e-06,
+      "loss": 0.2236,
+      "step": 1958
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.260828625235405e-06,
+      "loss": 0.2241,
+      "step": 1959
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.23728813559322e-06,
+      "loss": 0.2124,
+      "step": 1960
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.213747645951037e-06,
+      "loss": 0.2284,
+      "step": 1961
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.190207156308851e-06,
+      "loss": 0.2175,
+      "step": 1962
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 0.2305,
+      "step": 1963
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.143126177024482e-06,
+      "loss": 0.2059,
+      "step": 1964
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.119585687382298e-06,
+      "loss": 0.2249,
+      "step": 1965
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.096045197740113e-06,
+      "loss": 0.2425,
+      "step": 1966
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.072504708097929e-06,
+      "loss": 0.2217,
+      "step": 1967
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.048964218455744e-06,
+      "loss": 0.2336,
+      "step": 1968
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.025423728813559e-06,
+      "loss": 0.2269,
+      "step": 1969
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.001883239171375e-06,
+      "loss": 0.233,
+      "step": 1970
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.97834274952919e-06,
+      "loss": 0.2213,
+      "step": 1971
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.954802259887006e-06,
+      "loss": 0.2293,
+      "step": 1972
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.931261770244821e-06,
+      "loss": 0.224,
+      "step": 1973
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.907721280602637e-06,
+      "loss": 0.2306,
+      "step": 1974
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.884180790960452e-06,
+      "loss": 0.2348,
+      "step": 1975
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.860640301318268e-06,
+      "loss": 0.2251,
+      "step": 1976
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.8370998116760826e-06,
+      "loss": 0.241,
+      "step": 1977
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.813559322033899e-06,
+      "loss": 0.2276,
+      "step": 1978
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.790018832391714e-06,
+      "loss": 0.241,
+      "step": 1979
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.766478342749529e-06,
+      "loss": 0.2316,
+      "step": 1980
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.742937853107345e-06,
+      "loss": 0.2186,
+      "step": 1981
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.71939736346516e-06,
+      "loss": 0.239,
+      "step": 1982
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6958568738229753e-06,
+      "loss": 0.2256,
+      "step": 1983
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6723163841807913e-06,
+      "loss": 0.2307,
+      "step": 1984
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6487758945386064e-06,
+      "loss": 0.2427,
+      "step": 1985
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.625235404896422e-06,
+      "loss": 0.2289,
+      "step": 1986
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6016949152542374e-06,
+      "loss": 0.2263,
+      "step": 1987
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.578154425612053e-06,
+      "loss": 0.2295,
+      "step": 1988
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5546139359698685e-06,
+      "loss": 0.2289,
+      "step": 1989
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.531073446327684e-06,
+      "loss": 0.2254,
+      "step": 1990
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.507532956685499e-06,
+      "loss": 0.2265,
+      "step": 1991
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.483992467043315e-06,
+      "loss": 0.2337,
+      "step": 1992
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.46045197740113e-06,
+      "loss": 0.2176,
+      "step": 1993
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4369114877589453e-06,
+      "loss": 0.2367,
+      "step": 1994
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.413370998116761e-06,
+      "loss": 0.2284,
+      "step": 1995
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3898305084745763e-06,
+      "loss": 0.2346,
+      "step": 1996
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3662900188323914e-06,
+      "loss": 0.2305,
+      "step": 1997
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3427495291902074e-06,
+      "loss": 0.2211,
+      "step": 1998
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.319209039548023e-06,
+      "loss": 0.2202,
+      "step": 1999
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.295668549905838e-06,
+      "loss": 0.2607,
+      "step": 2000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.272128060263654e-06,
+      "loss": 0.2321,
+      "step": 2001
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.248587570621469e-06,
+      "loss": 0.2085,
+      "step": 2002
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.225047080979284e-06,
+      "loss": 0.2353,
+      "step": 2003
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2015065913371e-06,
+      "loss": 0.2229,
+      "step": 2004
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1779661016949152e-06,
+      "loss": 0.2193,
+      "step": 2005
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.154425612052731e-06,
+      "loss": 0.2292,
+      "step": 2006
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1308851224105463e-06,
+      "loss": 0.2322,
+      "step": 2007
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.107344632768362e-06,
+      "loss": 0.2286,
+      "step": 2008
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.083804143126177e-06,
+      "loss": 0.236,
+      "step": 2009
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0602636534839924e-06,
+      "loss": 0.2339,
+      "step": 2010
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0367231638418084e-06,
+      "loss": 0.2327,
+      "step": 2011
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0131826741996235e-06,
+      "loss": 0.2277,
+      "step": 2012
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.989642184557439e-06,
+      "loss": 0.2355,
+      "step": 2013
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9661016949152545e-06,
+      "loss": 0.2408,
+      "step": 2014
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9425612052730696e-06,
+      "loss": 0.2244,
+      "step": 2015
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.919020715630885e-06,
+      "loss": 0.2189,
+      "step": 2016
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8954802259887007e-06,
+      "loss": 0.2236,
+      "step": 2017
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8719397363465162e-06,
+      "loss": 0.2187,
+      "step": 2018
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8483992467043313e-06,
+      "loss": 0.2215,
+      "step": 2019
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.824858757062147e-06,
+      "loss": 0.2248,
+      "step": 2020
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8013182674199624e-06,
+      "loss": 0.2318,
+      "step": 2021
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.777777777777778e-06,
+      "loss": 0.2357,
+      "step": 2022
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7542372881355934e-06,
+      "loss": 0.247,
+      "step": 2023
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.730696798493409e-06,
+      "loss": 0.2299,
+      "step": 2024
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7071563088512245e-06,
+      "loss": 0.2305,
+      "step": 2025
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6836158192090396e-06,
+      "loss": 0.2246,
+      "step": 2026
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.660075329566855e-06,
+      "loss": 0.2387,
+      "step": 2027
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6365348399246707e-06,
+      "loss": 0.2212,
+      "step": 2028
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6129943502824858e-06,
+      "loss": 0.232,
+      "step": 2029
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.5894538606403013e-06,
+      "loss": 0.2371,
+      "step": 2030
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.565913370998117e-06,
+      "loss": 0.2312,
+      "step": 2031
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5423728813559323e-06,
+      "loss": 0.2328,
+      "step": 2032
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.518832391713748e-06,
+      "loss": 0.2291,
+      "step": 2033
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4952919020715634e-06,
+      "loss": 0.2304,
+      "step": 2034
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.471751412429379e-06,
+      "loss": 0.2384,
+      "step": 2035
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.448210922787194e-06,
+      "loss": 0.2224,
+      "step": 2036
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4246704331450096e-06,
+      "loss": 0.2259,
+      "step": 2037
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.401129943502825e-06,
+      "loss": 0.236,
+      "step": 2038
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.37758945386064e-06,
+      "loss": 0.2355,
+      "step": 2039
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3540489642184557e-06,
+      "loss": 0.2249,
+      "step": 2040
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3305084745762712e-06,
+      "loss": 0.2487,
+      "step": 2041
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3069679849340868e-06,
+      "loss": 0.2458,
+      "step": 2042
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.283427495291902e-06,
+      "loss": 0.2259,
+      "step": 2043
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.259887005649718e-06,
+      "loss": 0.2129,
+      "step": 2044
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2363465160075334e-06,
+      "loss": 0.2169,
+      "step": 2045
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2128060263653485e-06,
+      "loss": 0.2306,
+      "step": 2046
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.189265536723164e-06,
+      "loss": 0.2235,
+      "step": 2047
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1657250470809795e-06,
+      "loss": 0.2399,
+      "step": 2048
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.142184557438795e-06,
+      "loss": 0.2502,
+      "step": 2049
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.11864406779661e-06,
+      "loss": 0.2242,
+      "step": 2050
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0951035781544257e-06,
+      "loss": 0.245,
+      "step": 2051
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.071563088512241e-06,
+      "loss": 0.2342,
+      "step": 2052
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0480225988700563e-06,
+      "loss": 0.2257,
+      "step": 2053
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.024482109227872e-06,
+      "loss": 0.2278,
+      "step": 2054
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0009416195856874e-06,
+      "loss": 0.2215,
+      "step": 2055
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.977401129943503e-06,
+      "loss": 0.2167,
+      "step": 2056
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9538606403013184e-06,
+      "loss": 0.2371,
+      "step": 2057
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.930320150659134e-06,
+      "loss": 0.2384,
+      "step": 2058
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9067796610169495e-06,
+      "loss": 0.2222,
+      "step": 2059
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8832391713747646e-06,
+      "loss": 0.232,
+      "step": 2060
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.85969868173258e-06,
+      "loss": 0.2356,
+      "step": 2061
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8361581920903956e-06,
+      "loss": 0.2242,
+      "step": 2062
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.812617702448211e-06,
+      "loss": 0.2176,
+      "step": 2063
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7890772128060265e-06,
+      "loss": 0.2186,
+      "step": 2064
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.765536723163842e-06,
+      "loss": 0.209,
+      "step": 2065
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7419962335216575e-06,
+      "loss": 0.2359,
+      "step": 2066
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7184557438794726e-06,
+      "loss": 0.2391,
+      "step": 2067
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6949152542372882e-06,
+      "loss": 0.2295,
+      "step": 2068
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6713747645951037e-06,
+      "loss": 0.2376,
+      "step": 2069
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.647834274952919e-06,
+      "loss": 0.2522,
+      "step": 2070
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6242937853107345e-06,
+      "loss": 0.2329,
+      "step": 2071
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.60075329566855e-06,
+      "loss": 0.2394,
+      "step": 2072
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5772128060263656e-06,
+      "loss": 0.2314,
+      "step": 2073
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.553672316384181e-06,
+      "loss": 0.2281,
+      "step": 2074
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5301318267419962e-06,
+      "loss": 0.2477,
+      "step": 2075
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5065913370998117e-06,
+      "loss": 0.2386,
+      "step": 2076
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4830508474576273e-06,
+      "loss": 0.2217,
+      "step": 2077
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4595103578154426e-06,
+      "loss": 0.2238,
+      "step": 2078
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4359698681732581e-06,
+      "loss": 0.2212,
+      "step": 2079
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4124293785310734e-06,
+      "loss": 0.2349,
+      "step": 2080
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.388888888888889e-06,
+      "loss": 0.2311,
+      "step": 2081
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3653483992467045e-06,
+      "loss": 0.2269,
+      "step": 2082
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3418079096045198e-06,
+      "loss": 0.2308,
+      "step": 2083
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3182674199623353e-06,
+      "loss": 0.2133,
+      "step": 2084
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2947269303201506e-06,
+      "loss": 0.24,
+      "step": 2085
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2711864406779662e-06,
+      "loss": 0.2233,
+      "step": 2086
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2476459510357817e-06,
+      "loss": 0.2342,
+      "step": 2087
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.224105461393597e-06,
+      "loss": 0.2424,
+      "step": 2088
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2005649717514125e-06,
+      "loss": 0.2169,
+      "step": 2089
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1770244821092279e-06,
+      "loss": 0.2111,
+      "step": 2090
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1534839924670434e-06,
+      "loss": 0.2242,
+      "step": 2091
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.129943502824859e-06,
+      "loss": 0.229,
+      "step": 2092
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1064030131826742e-06,
+      "loss": 0.2274,
+      "step": 2093
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.0828625235404898e-06,
+      "loss": 0.2291,
+      "step": 2094
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.059322033898305e-06,
+      "loss": 0.2257,
+      "step": 2095
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0357815442561206e-06,
+      "loss": 0.2102,
+      "step": 2096
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.012241054613936e-06,
+      "loss": 0.2343,
+      "step": 2097
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.887005649717514e-07,
+      "loss": 0.23,
+      "step": 2098
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.65160075329567e-07,
+      "loss": 0.2353,
+      "step": 2099
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.416195856873823e-07,
+      "loss": 0.2275,
+      "step": 2100
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.180790960451978e-07,
+      "loss": 0.219,
+      "step": 2101
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.945386064030132e-07,
+      "loss": 0.229,
+      "step": 2102
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.709981167608288e-07,
+      "loss": 0.2162,
+      "step": 2103
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.474576271186441e-07,
+      "loss": 0.2289,
+      "step": 2104
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.239171374764595e-07,
+      "loss": 0.2263,
+      "step": 2105
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.00376647834275e-07,
+      "loss": 0.2183,
+      "step": 2106
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.768361581920904e-07,
+      "loss": 0.224,
+      "step": 2107
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.532956685499059e-07,
+      "loss": 0.2263,
+      "step": 2108
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.297551789077213e-07,
+      "loss": 0.2277,
+      "step": 2109
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.062146892655367e-07,
+      "loss": 0.2174,
+      "step": 2110
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.826741996233522e-07,
+      "loss": 0.2242,
+      "step": 2111
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.591337099811677e-07,
+      "loss": 0.2293,
+      "step": 2112
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.355932203389831e-07,
+      "loss": 0.24,
+      "step": 2113
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.120527306967985e-07,
+      "loss": 0.2288,
+      "step": 2114
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.885122410546139e-07,
+      "loss": 0.2228,
+      "step": 2115
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.649717514124295e-07,
+      "loss": 0.2404,
+      "step": 2116
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.414312617702449e-07,
+      "loss": 0.2235,
+      "step": 2117
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.178907721280603e-07,
+      "loss": 0.2289,
+      "step": 2118
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.943502824858757e-07,
+      "loss": 0.2281,
+      "step": 2119
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.7080979284369114e-07,
+      "loss": 0.2234,
+      "step": 2120
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.472693032015066e-07,
+      "loss": 0.2251,
+      "step": 2121
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.2372881355932204e-07,
+      "loss": 0.23,
+      "step": 2122
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.001883239171375e-07,
+      "loss": 0.2219,
+      "step": 2123
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.7664783427495294e-07,
+      "loss": 0.2282,
+      "step": 2124
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5310734463276836e-07,
+      "loss": 0.2148,
+      "step": 2125
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2956685499058383e-07,
+      "loss": 0.2332,
+      "step": 2126
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.0602636534839925e-07,
+      "loss": 0.2233,
+      "step": 2127
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.8248587570621473e-07,
+      "loss": 0.2364,
+      "step": 2128
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5894538606403015e-07,
+      "loss": 0.2215,
+      "step": 2129
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3540489642184557e-07,
+      "loss": 0.2266,
+      "step": 2130
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.1186440677966102e-07,
+      "loss": 0.2283,
+      "step": 2131
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8832391713747647e-07,
+      "loss": 0.2406,
+      "step": 2132
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.6478342749529192e-07,
+      "loss": 0.2148,
+      "step": 2133
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4124293785310736e-07,
+      "loss": 0.2292,
+      "step": 2134
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1770244821092279e-07,
+      "loss": 0.2232,
+      "step": 2135
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.416195856873823e-08,
+      "loss": 0.2214,
+      "step": 2136
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.062146892655368e-08,
+      "loss": 0.2167,
+      "step": 2137
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.708097928436912e-08,
+      "loss": 0.237,
+      "step": 2138
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.354048964218456e-08,
+      "loss": 0.234,
+      "step": 2139
+    },
+    {
+      "epoch": 3.0,
+      "step": 2139,
+      "total_flos": 503763904823296.0,
+      "train_loss": 0.15229338424932518,
+      "train_runtime": 92260.4068,
+      "train_samples_per_second": 5.942,
+      "train_steps_per_second": 0.023
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2139,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "total_flos": 503763904823296.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}