diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,15955 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9983060417843026,
+  "global_step": 2655,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5000000000000004e-07,
+      "loss": 2.5126,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 2.5546,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.5e-07,
+      "loss": 2.5761,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 2.4336,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.25e-06,
+      "loss": 2.4224,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5e-06,
+      "loss": 2.5806,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.75e-06,
+      "loss": 2.4375,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 2.3364,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.25e-06,
+      "loss": 2.4234,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-06,
+      "loss": 2.4857,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7500000000000004e-06,
+      "loss": 2.4034,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-06,
+      "loss": 2.3313,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2500000000000002e-06,
+      "loss": 2.3999,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5e-06,
+      "loss": 2.2607,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 2.4289,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 2.2785,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.25e-06,
+      "loss": 2.3179,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5e-06,
+      "loss": 2.1615,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.75e-06,
+      "loss": 2.2085,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-06,
+      "loss": 2.4545,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2500000000000006e-06,
+      "loss": 2.3324,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 2.2629,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.75e-06,
+      "loss": 2.3195,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6e-06,
+      "loss": 2.2893,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.25e-06,
+      "loss": 1.9683,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 2.2137,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.750000000000001e-06,
+      "loss": 2.1238,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7e-06,
+      "loss": 2.0576,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.25e-06,
+      "loss": 2.2156,
+      "step": 29
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 2.2378,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.75e-06,
+      "loss": 2.0567,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 2.0553,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.25e-06,
+      "loss": 2.0463,
+      "step": 33
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.5e-06,
+      "loss": 2.0772,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.750000000000001e-06,
+      "loss": 2.0675,
+      "step": 35
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9e-06,
+      "loss": 1.8712,
+      "step": 36
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.250000000000001e-06,
+      "loss": 1.8788,
+      "step": 37
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.5e-06,
+      "loss": 2.006,
+      "step": 38
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.75e-06,
+      "loss": 1.9514,
+      "step": 39
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1e-05,
+      "loss": 1.7712,
+      "step": 40
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.025e-05,
+      "loss": 1.9156,
+      "step": 41
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0500000000000001e-05,
+      "loss": 1.7753,
+      "step": 42
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.075e-05,
+      "loss": 1.8338,
+      "step": 43
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 1.7189,
+      "step": 44
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.125e-05,
+      "loss": 1.8162,
+      "step": 45
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.15e-05,
+      "loss": 1.7132,
+      "step": 46
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1750000000000001e-05,
+      "loss": 1.7059,
+      "step": 47
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2e-05,
+      "loss": 1.8452,
+      "step": 48
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2250000000000001e-05,
+      "loss": 1.672,
+      "step": 49
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.25e-05,
+      "loss": 1.5752,
+      "step": 50
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.275e-05,
+      "loss": 1.6296,
+      "step": 51
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 1.6531,
+      "step": 52
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.325e-05,
+      "loss": 1.7057,
+      "step": 53
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 1.605,
+      "step": 54
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.375e-05,
+      "loss": 1.6354,
+      "step": 55
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4e-05,
+      "loss": 1.6594,
+      "step": 56
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.425e-05,
+      "loss": 1.5898,
+      "step": 57
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.45e-05,
+      "loss": 1.4594,
+      "step": 58
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4750000000000003e-05,
+      "loss": 1.6012,
+      "step": 59
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 1.4589,
+      "step": 60
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.525e-05,
+      "loss": 1.5964,
+      "step": 61
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.55e-05,
+      "loss": 1.4511,
+      "step": 62
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.575e-05,
+      "loss": 1.4865,
+      "step": 63
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.4449,
+      "step": 64
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6250000000000002e-05,
+      "loss": 1.4099,
+      "step": 65
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.65e-05,
+      "loss": 1.4373,
+      "step": 66
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.675e-05,
+      "loss": 1.1847,
+      "step": 67
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7e-05,
+      "loss": 1.2191,
+      "step": 68
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7250000000000003e-05,
+      "loss": 1.348,
+      "step": 69
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 1.2346,
+      "step": 70
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.775e-05,
+      "loss": 1.3182,
+      "step": 71
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8e-05,
+      "loss": 1.3917,
+      "step": 72
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.825e-05,
+      "loss": 1.1922,
+      "step": 73
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8500000000000002e-05,
+      "loss": 1.223,
+      "step": 74
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 1.1925,
+      "step": 75
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9e-05,
+      "loss": 1.2361,
+      "step": 76
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.925e-05,
+      "loss": 1.2043,
+      "step": 77
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.95e-05,
+      "loss": 1.2393,
+      "step": 78
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9750000000000002e-05,
+      "loss": 1.2596,
+      "step": 79
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 1.2369,
+      "step": 80
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999992557561938e-05,
+      "loss": 1.1721,
+      "step": 81
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999997023025883e-05,
+      "loss": 1.148,
+      "step": 82
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999933018123898e-05,
+      "loss": 1.0804,
+      "step": 83
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999880921212545e-05,
+      "loss": 1.2189,
+      "step": 84
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999813939602312e-05,
+      "loss": 1.0859,
+      "step": 85
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999732073392906e-05,
+      "loss": 1.1502,
+      "step": 86
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999635322706173e-05,
+      "loss": 1.2196,
+      "step": 87
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999952368768613e-05,
+      "loss": 1.05,
+      "step": 88
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999939716849895e-05,
+      "loss": 1.1159,
+      "step": 89
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999255765332947e-05,
+      "loss": 1.0871,
+      "step": 90
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.99990994783986e-05,
+      "loss": 1.0078,
+      "step": 91
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998928307928544e-05,
+      "loss": 1.0909,
+      "step": 92
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998742254177562e-05,
+      "loss": 1.0635,
+      "step": 93
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998541317422592e-05,
+      "loss": 1.0716,
+      "step": 94
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998325497962724e-05,
+      "loss": 1.1236,
+      "step": 95
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998094796119204e-05,
+      "loss": 1.0877,
+      "step": 96
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997849212235432e-05,
+      "loss": 1.0257,
+      "step": 97
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997588746676955e-05,
+      "loss": 1.1584,
+      "step": 98
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997313399831472e-05,
+      "loss": 1.0707,
+      "step": 99
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999702317210883e-05,
+      "loss": 1.0438,
+      "step": 100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996718063941034e-05,
+      "loss": 1.0072,
+      "step": 101
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996398075782234e-05,
+      "loss": 0.986,
+      "step": 102
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996063208108723e-05,
+      "loss": 0.9915,
+      "step": 103
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995713461418953e-05,
+      "loss": 1.0452,
+      "step": 104
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995348836233517e-05,
+      "loss": 0.9564,
+      "step": 105
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994969333095154e-05,
+      "loss": 0.8627,
+      "step": 106
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999457495256875e-05,
+      "loss": 1.0193,
+      "step": 107
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999416569524133e-05,
+      "loss": 1.0176,
+      "step": 108
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993741561722076e-05,
+      "loss": 0.9931,
+      "step": 109
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993302552642306e-05,
+      "loss": 1.0409,
+      "step": 110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992848668655473e-05,
+      "loss": 0.972,
+      "step": 111
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992379910437183e-05,
+      "loss": 1.0317,
+      "step": 112
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991896278685176e-05,
+      "loss": 0.9879,
+      "step": 113
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999139777411933e-05,
+      "loss": 0.9572,
+      "step": 114
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990884397481664e-05,
+      "loss": 0.9143,
+      "step": 115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990356149536332e-05,
+      "loss": 1.0724,
+      "step": 116
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989813031069624e-05,
+      "loss": 0.943,
+      "step": 117
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998925504288997e-05,
+      "loss": 0.9552,
+      "step": 118
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998868218582792e-05,
+      "loss": 1.0169,
+      "step": 119
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988094460736175e-05,
+      "loss": 0.9943,
+      "step": 120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987491868489545e-05,
+      "loss": 1.0235,
+      "step": 121
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986874409984992e-05,
+      "loss": 0.9166,
+      "step": 122
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986242086141584e-05,
+      "loss": 1.0185,
+      "step": 123
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998559489790054e-05,
+      "loss": 0.9538,
+      "step": 124
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998493284622518e-05,
+      "loss": 1.0077,
+      "step": 125
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998425593210097e-05,
+      "loss": 0.9701,
+      "step": 126
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998356415653548e-05,
+      "loss": 0.8868,
+      "step": 127
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982857520558413e-05,
+      "loss": 0.9739,
+      "step": 128
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982136025221587e-05,
+      "loss": 0.9177,
+      "step": 129
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998139967159894e-05,
+      "loss": 0.972,
+      "step": 130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980648460786524e-05,
+      "loss": 0.9609,
+      "step": 131
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979882393902507e-05,
+      "loss": 0.9346,
+      "step": 132
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979101472087175e-05,
+      "loss": 1.0007,
+      "step": 133
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978305696502912e-05,
+      "loss": 0.8533,
+      "step": 134
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977495068334223e-05,
+      "loss": 0.8527,
+      "step": 135
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976669588787723e-05,
+      "loss": 0.9034,
+      "step": 136
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997582925909212e-05,
+      "loss": 0.9523,
+      "step": 137
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997497408049824e-05,
+      "loss": 0.9795,
+      "step": 138
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974104054279006e-05,
+      "loss": 0.8871,
+      "step": 139
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997321918172944e-05,
+      "loss": 0.9368,
+      "step": 140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972319464166663e-05,
+      "loss": 0.8211,
+      "step": 141
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971404902929892e-05,
+      "loss": 0.8928,
+      "step": 142
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970475499380444e-05,
+      "loss": 0.8974,
+      "step": 143
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996953125490172e-05,
+      "loss": 0.8924,
+      "step": 144
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996857217089922e-05,
+      "loss": 0.8958,
+      "step": 145
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996759824880053e-05,
+      "loss": 0.888,
+      "step": 146
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966609490055314e-05,
+      "loss": 0.8902,
+      "step": 147
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965605896135336e-05,
+      "loss": 0.8939,
+      "step": 148
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964587468534428e-05,
+      "loss": 0.8764,
+      "step": 149
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963554208768502e-05,
+      "loss": 0.8629,
+      "step": 150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962506118375565e-05,
+      "loss": 0.9657,
+      "step": 151
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996144319891568e-05,
+      "loss": 0.8994,
+      "step": 152
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996036545197098e-05,
+      "loss": 0.9128,
+      "step": 153
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995927287914569e-05,
+      "loss": 0.9005,
+      "step": 154
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9958165482066094e-05,
+      "loss": 0.9554,
+      "step": 155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995704326238053e-05,
+      "loss": 0.8606,
+      "step": 156
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955906221759412e-05,
+      "loss": 0.9079,
+      "step": 157
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954754361895204e-05,
+      "loss": 0.8207,
+      "step": 158
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995358768450245e-05,
+      "loss": 0.8818,
+      "step": 159
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952406191317718e-05,
+      "loss": 0.8702,
+      "step": 160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995120988409966e-05,
+      "loss": 0.918,
+      "step": 161
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949998764628952e-05,
+      "loss": 0.8607,
+      "step": 162
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994877283470834e-05,
+      "loss": 0.8258,
+      "step": 163
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9947532096162603e-05,
+      "loss": 1.0383,
+      "step": 164
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994627655083856e-05,
+      "loss": 0.7616,
+      "step": 165
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945006200605082e-05,
+      "loss": 0.8759,
+      "step": 166
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994372104735307e-05,
+      "loss": 0.8444,
+      "step": 167
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994242109299545e-05,
+      "loss": 0.8571,
+      "step": 168
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941106339467194e-05,
+      "loss": 0.9131,
+      "step": 169
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939776788725296e-05,
+      "loss": 0.9078,
+      "step": 170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938432442748773e-05,
+      "loss": 0.8503,
+      "step": 171
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937073303538673e-05,
+      "loss": 0.779,
+      "step": 172
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993569937311805e-05,
+      "loss": 0.8212,
+      "step": 173
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993431065353199e-05,
+      "loss": 0.7926,
+      "step": 174
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993290714684758e-05,
+      "loss": 0.8726,
+      "step": 175
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931488855153927e-05,
+      "loss": 0.8757,
+      "step": 176
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930055780562133e-05,
+      "loss": 0.8542,
+      "step": 177
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992860792520532e-05,
+      "loss": 0.9084,
+      "step": 178
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99271452912386e-05,
+      "loss": 0.8381,
+      "step": 179
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992566788083908e-05,
+      "loss": 0.8779,
+      "step": 180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924175696205873e-05,
+      "loss": 0.8883,
+      "step": 181
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992266873956008e-05,
+      "loss": 0.7767,
+      "step": 182
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921147013144782e-05,
+      "loss": 0.8474,
+      "step": 183
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919610519225046e-05,
+      "loss": 0.8741,
+      "step": 184
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918059260087937e-05,
+      "loss": 0.7962,
+      "step": 185
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916493238042475e-05,
+      "loss": 0.8783,
+      "step": 186
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914912455419662e-05,
+      "loss": 0.8591,
+      "step": 187
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913316914572483e-05,
+      "loss": 0.8354,
+      "step": 188
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911706617875875e-05,
+      "loss": 0.8882,
+      "step": 189
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910081567726746e-05,
+      "loss": 0.8215,
+      "step": 190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908441766543963e-05,
+      "loss": 0.7782,
+      "step": 191
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9906787216768347e-05,
+      "loss": 0.888,
+      "step": 192
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905117920862684e-05,
+      "loss": 0.8782,
+      "step": 193
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903433881311686e-05,
+      "loss": 0.8781,
+      "step": 194
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901735100622038e-05,
+      "loss": 0.7871,
+      "step": 195
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900021581322347e-05,
+      "loss": 0.8192,
+      "step": 196
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898293325963164e-05,
+      "loss": 0.8503,
+      "step": 197
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896550337116984e-05,
+      "loss": 0.7527,
+      "step": 198
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9894792617378215e-05,
+      "loss": 0.8162,
+      "step": 199
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893020169363203e-05,
+      "loss": 0.9062,
+      "step": 200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9891232995710222e-05,
+      "loss": 0.8545,
+      "step": 201
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988943109907945e-05,
+      "loss": 0.7395,
+      "step": 202
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988761448215299e-05,
+      "loss": 0.8984,
+      "step": 203
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885783147634856e-05,
+      "loss": 0.8559,
+      "step": 204
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883937098250962e-05,
+      "loss": 0.8422,
+      "step": 205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882076336749137e-05,
+      "loss": 0.7944,
+      "step": 206
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988020086589909e-05,
+      "loss": 0.8565,
+      "step": 207
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9878310688492452e-05,
+      "loss": 0.8417,
+      "step": 208
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876405807342717e-05,
+      "loss": 0.8349,
+      "step": 209
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9874486225285278e-05,
+      "loss": 0.7939,
+      "step": 210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872551945177414e-05,
+      "loss": 0.8898,
+      "step": 211
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870602969898273e-05,
+      "loss": 0.8254,
+      "step": 212
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986863930234888e-05,
+      "loss": 0.7909,
+      "step": 213
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866660945452133e-05,
+      "loss": 0.7217,
+      "step": 214
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986466790215279e-05,
+      "loss": 0.8383,
+      "step": 215
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9862660175417467e-05,
+      "loss": 0.7865,
+      "step": 216
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.986063776823465e-05,
+      "loss": 0.8801,
+      "step": 217
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985860068361466e-05,
+      "loss": 0.8641,
+      "step": 218
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9856548924589673e-05,
+      "loss": 0.8206,
+      "step": 219
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985448249421371e-05,
+      "loss": 0.8147,
+      "step": 220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9852401395562624e-05,
+      "loss": 0.8406,
+      "step": 221
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9850305631734105e-05,
+      "loss": 0.8049,
+      "step": 222
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848195205847672e-05,
+      "loss": 0.7941,
+      "step": 223
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846070121044668e-05,
+      "loss": 0.7481,
+      "step": 224
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9843930380488257e-05,
+      "loss": 0.9365,
+      "step": 225
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9841775987363415e-05,
+      "loss": 0.858,
+      "step": 226
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9839606944876927e-05,
+      "loss": 0.8136,
+      "step": 227
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9837423256257388e-05,
+      "loss": 0.8204,
+      "step": 228
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983522492475519e-05,
+      "loss": 0.8449,
+      "step": 229
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9833011953642525e-05,
+      "loss": 0.7347,
+      "step": 230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9830784346213373e-05,
+      "loss": 0.8719,
+      "step": 231
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98285421057835e-05,
+      "loss": 0.7999,
+      "step": 232
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9826285235690447e-05,
+      "loss": 0.8624,
+      "step": 233
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9824013739293542e-05,
+      "loss": 0.7819,
+      "step": 234
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9821727619973884e-05,
+      "loss": 0.8069,
+      "step": 235
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9819426881134324e-05,
+      "loss": 0.7801,
+      "step": 236
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981711152619949e-05,
+      "loss": 0.7876,
+      "step": 237
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9814781558615755e-05,
+      "loss": 0.8151,
+      "step": 238
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981243698185125e-05,
+      "loss": 0.8107,
+      "step": 239
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9810077799395847e-05,
+      "loss": 0.7613,
+      "step": 240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9807704014761163e-05,
+      "loss": 0.8426,
+      "step": 241
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980531563148054e-05,
+      "loss": 0.7713,
+      "step": 242
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9802912653109063e-05,
+      "loss": 0.8683,
+      "step": 243
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9800495083223536e-05,
+      "loss": 0.8024,
+      "step": 244
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9798062925422474e-05,
+      "loss": 0.7895,
+      "step": 245
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9795616183326124e-05,
+      "loss": 0.7833,
+      "step": 246
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9793154860576424e-05,
+      "loss": 0.7771,
+      "step": 247
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9790678960837028e-05,
+      "loss": 0.8281,
+      "step": 248
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9788188487793273e-05,
+      "loss": 0.7862,
+      "step": 249
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9785683445152205e-05,
+      "loss": 0.8251,
+      "step": 250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9783163836642553e-05,
+      "loss": 0.8754,
+      "step": 251
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9780629666014713e-05,
+      "loss": 0.8352,
+      "step": 252
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977808093704077e-05,
+      "loss": 0.8285,
+      "step": 253
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9775517653514478e-05,
+      "loss": 0.835,
+      "step": 254
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9772939819251247e-05,
+      "loss": 0.7272,
+      "step": 255
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9770347438088162e-05,
+      "loss": 0.7905,
+      "step": 256
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9767740513883942e-05,
+      "loss": 0.7938,
+      "step": 257
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9765119050518963e-05,
+      "loss": 0.7838,
+      "step": 258
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976248305189524e-05,
+      "loss": 0.8081,
+      "step": 259
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9759832521936424e-05,
+      "loss": 0.8196,
+      "step": 260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.97571674645878e-05,
+      "loss": 0.7775,
+      "step": 261
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9754487883816268e-05,
+      "loss": 0.8208,
+      "step": 262
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9751793783610353e-05,
+      "loss": 0.8199,
+      "step": 263
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9749085167980192e-05,
+      "loss": 0.7553,
+      "step": 264
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974636204095752e-05,
+      "loss": 0.8164,
+      "step": 265
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9743624406595684e-05,
+      "loss": 0.7751,
+      "step": 266
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9740872268969613e-05,
+      "loss": 0.7183,
+      "step": 267
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9738105632175837e-05,
+      "loss": 0.7244,
+      "step": 268
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9735324500332455e-05,
+      "loss": 0.7686,
+      "step": 269
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9732528877579145e-05,
+      "loss": 0.7869,
+      "step": 270
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9729718768077165e-05,
+      "loss": 0.7908,
+      "step": 271
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9726894176009322e-05,
+      "loss": 0.7805,
+      "step": 272
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972405510557999e-05,
+      "loss": 0.7435,
+      "step": 273
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9721201561015088e-05,
+      "loss": 0.8312,
+      "step": 274
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971833354656208e-05,
+      "loss": 0.7819,
+      "step": 275
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9715451066489973e-05,
+      "loss": 0.8309,
+      "step": 276
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9712554125089303e-05,
+      "loss": 0.869,
+      "step": 277
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970964272667213e-05,
+      "loss": 0.7434,
+      "step": 278
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970671687557203e-05,
+      "loss": 0.7118,
+      "step": 279
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9703776576144106e-05,
+      "loss": 0.8012,
+      "step": 280
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9700821832764952e-05,
+      "loss": 0.7813,
+      "step": 281
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9697852649832668e-05,
+      "loss": 0.9035,
+      "step": 282
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969486903176684e-05,
+      "loss": 0.7834,
+      "step": 283
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9691870983008557e-05,
+      "loss": 0.8584,
+      "step": 284
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968885850802037e-05,
+      "loss": 0.7233,
+      "step": 285
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9685831611286312e-05,
+      "loss": 0.7469,
+      "step": 286
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968279029731188e-05,
+      "loss": 0.7824,
+      "step": 287
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967973457062404e-05,
+      "loss": 0.782,
+      "step": 288
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9676664435771196e-05,
+      "loss": 0.6867,
+      "step": 289
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9673579897323204e-05,
+      "loss": 0.7786,
+      "step": 290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967048095987137e-05,
+      "loss": 0.8254,
+      "step": 291
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9667367628028412e-05,
+      "loss": 0.7814,
+      "step": 292
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9664239906428494e-05,
+      "loss": 0.8722,
+      "step": 293
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9661097799727193e-05,
+      "loss": 0.7607,
+      "step": 294
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9657941312601486e-05,
+      "loss": 0.8115,
+      "step": 295
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9654770449749777e-05,
+      "loss": 0.8139,
+      "step": 296
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9651585215891845e-05,
+      "loss": 0.7655,
+      "step": 297
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9648385615768882e-05,
+      "loss": 0.7921,
+      "step": 298
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9645171654143444e-05,
+      "loss": 0.7659,
+      "step": 299
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964194333579948e-05,
+      "loss": 0.8443,
+      "step": 300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9638700665542293e-05,
+      "loss": 0.7309,
+      "step": 301
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9635443648198566e-05,
+      "loss": 0.8203,
+      "step": 302
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9632172288616328e-05,
+      "loss": 0.7384,
+      "step": 303
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9628886591664954e-05,
+      "loss": 0.7675,
+      "step": 304
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962558656223516e-05,
+      "loss": 0.7853,
+      "step": 305
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9622272205239007e-05,
+      "loss": 0.6857,
+      "step": 306
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9618943525609874e-05,
+      "loss": 0.8464,
+      "step": 307
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961560052830245e-05,
+      "loss": 0.6699,
+      "step": 308
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9612243218292753e-05,
+      "loss": 0.8644,
+      "step": 309
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9608871600578095e-05,
+      "loss": 0.7354,
+      "step": 310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960548568017709e-05,
+      "loss": 0.7987,
+      "step": 311
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9602085462129635e-05,
+      "loss": 0.7828,
+      "step": 312
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959867095149691e-05,
+      "loss": 0.7492,
+      "step": 313
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959524215336138e-05,
+      "loss": 0.7494,
+      "step": 314
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9591799072826766e-05,
+      "loss": 0.7713,
+      "step": 315
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9588341715018043e-05,
+      "loss": 0.6961,
+      "step": 316
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958487008508145e-05,
+      "loss": 0.7363,
+      "step": 317
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9581384188184475e-05,
+      "loss": 0.7725,
+      "step": 318
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9577884029515816e-05,
+      "loss": 0.7178,
+      "step": 319
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9574369614285426e-05,
+      "loss": 0.7621,
+      "step": 320
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9570840947724465e-05,
+      "loss": 0.8003,
+      "step": 321
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956729803508531e-05,
+      "loss": 0.8216,
+      "step": 322
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9563740881641548e-05,
+      "loss": 0.6778,
+      "step": 323
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9560169492687947e-05,
+      "loss": 0.7491,
+      "step": 324
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9556583873540483e-05,
+      "loss": 0.7385,
+      "step": 325
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95529840295363e-05,
+      "loss": 0.8002,
+      "step": 326
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954936996603373e-05,
+      "loss": 0.8454,
+      "step": 327
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9545741688412256e-05,
+      "loss": 0.7519,
+      "step": 328
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954209920207252e-05,
+      "loss": 0.7457,
+      "step": 329
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953844251243633e-05,
+      "loss": 0.6901,
+      "step": 330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953477162494661e-05,
+      "loss": 0.7058,
+      "step": 331
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9531086545067438e-05,
+      "loss": 0.7242,
+      "step": 332
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9527387278284008e-05,
+      "loss": 0.7541,
+      "step": 333
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9523673830102636e-05,
+      "loss": 0.7663,
+      "step": 334
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9519946206050737e-05,
+      "loss": 0.7698,
+      "step": 335
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951620441167684e-05,
+      "loss": 0.6764,
+      "step": 336
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9512448452550555e-05,
+      "loss": 0.7944,
+      "step": 337
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950867833426258e-05,
+      "loss": 0.8507,
+      "step": 338
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9504894062424696e-05,
+      "loss": 0.7292,
+      "step": 339
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9501095642669737e-05,
+      "loss": 0.7331,
+      "step": 340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9497283080651608e-05,
+      "loss": 0.7848,
+      "step": 341
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9493456382045258e-05,
+      "loss": 0.6951,
+      "step": 342
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9489615552546685e-05,
+      "loss": 0.7483,
+      "step": 343
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948576059787291e-05,
+      "loss": 0.7067,
+      "step": 344
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9481891523761985e-05,
+      "loss": 0.8267,
+      "step": 345
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947800833597299e-05,
+      "loss": 0.7235,
+      "step": 346
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947411104028599e-05,
+      "loss": 0.7498,
+      "step": 347
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9470199642502062e-05,
+      "loss": 0.6908,
+      "step": 348
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9466274148443282e-05,
+      "loss": 0.6926,
+      "step": 349
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.946233456395269e-05,
+      "loss": 0.7373,
+      "step": 350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9458380894894312e-05,
+      "loss": 0.7747,
+      "step": 351
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9454413147153137e-05,
+      "loss": 0.7443,
+      "step": 352
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945043132663511e-05,
+      "loss": 0.771,
+      "step": 353
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9446435439267115e-05,
+      "loss": 0.7553,
+      "step": 354
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9442425490996987e-05,
+      "loss": 0.7188,
+      "step": 355
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9438401487793484e-05,
+      "loss": 0.7377,
+      "step": 356
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9434363435646277e-05,
+      "loss": 0.7671,
+      "step": 357
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9430311340565967e-05,
+      "loss": 0.7072,
+      "step": 358
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942624520858404e-05,
+      "loss": 0.7085,
+      "step": 359
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942216504575289e-05,
+      "loss": 0.7896,
+      "step": 360
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941807085814578e-05,
+      "loss": 0.7702,
+      "step": 361
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9413962651856865e-05,
+      "loss": 0.8157,
+      "step": 362
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9409840433001153e-05,
+      "loss": 0.6909,
+      "step": 363
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9405704207714526e-05,
+      "loss": 0.7064,
+      "step": 364
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940155398215369e-05,
+      "loss": 0.7011,
+      "step": 365
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9397389762496218e-05,
+      "loss": 0.7073,
+      "step": 366
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939321155494049e-05,
+      "loss": 0.7593,
+      "step": 367
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9389019365705718e-05,
+      "loss": 0.7447,
+      "step": 368
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9384813201031923e-05,
+      "loss": 0.714,
+      "step": 369
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9380593067179934e-05,
+      "loss": 0.7854,
+      "step": 370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9376358970431365e-05,
+      "loss": 0.8076,
+      "step": 371
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9372110917088614e-05,
+      "loss": 0.7411,
+      "step": 372
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936784891347486e-05,
+      "loss": 0.7903,
+      "step": 373
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9363572965934044e-05,
+      "loss": 0.7735,
+      "step": 374
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9359283080830856e-05,
+      "loss": 0.6893,
+      "step": 375
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9354979264550734e-05,
+      "loss": 0.8029,
+      "step": 376
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9350661523499864e-05,
+      "loss": 0.6653,
+      "step": 377
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9346329864105144e-05,
+      "loss": 0.7215,
+      "step": 378
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.93419842928142e-05,
+      "loss": 0.7231,
+      "step": 379
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933762481609536e-05,
+      "loss": 0.8041,
+      "step": 380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933325144043764e-05,
+      "loss": 0.6891,
+      "step": 381
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9328864172350773e-05,
+      "loss": 0.7578,
+      "step": 382
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932446301836514e-05,
+      "loss": 0.7918,
+      "step": 383
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9320047985031812e-05,
+      "loss": 0.7536,
+      "step": 384
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9315619078922512e-05,
+      "loss": 0.7721,
+      "step": 385
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9311176306629603e-05,
+      "loss": 0.733,
+      "step": 386
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9306719674766105e-05,
+      "loss": 0.7631,
+      "step": 387
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9302249189965655e-05,
+      "loss": 0.7313,
+      "step": 388
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9297764858882516e-05,
+      "loss": 0.7616,
+      "step": 389
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929326668819156e-05,
+      "loss": 0.6834,
+      "step": 390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9288754684588252e-05,
+      "loss": 0.7014,
+      "step": 391
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9284228854788666e-05,
+      "loss": 0.7689,
+      "step": 392
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9279689205529432e-05,
+      "loss": 0.7665,
+      "step": 393
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9275135743567768e-05,
+      "loss": 0.7036,
+      "step": 394
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9270568475681442e-05,
+      "loss": 0.7918,
+      "step": 395
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926598740866878e-05,
+      "loss": 0.733,
+      "step": 396
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926139254934864e-05,
+      "loss": 0.763,
+      "step": 397
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925678390456041e-05,
+      "loss": 0.7451,
+      "step": 398
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925216148116401e-05,
+      "loss": 0.7278,
+      "step": 399
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9247525286039855e-05,
+      "loss": 0.715,
+      "step": 400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924287532608886e-05,
+      "loss": 0.6863,
+      "step": 401
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9238211608232436e-05,
+      "loss": 0.6509,
+      "step": 402
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9233534139412473e-05,
+      "loss": 0.7453,
+      "step": 403
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9228842926591323e-05,
+      "loss": 0.7919,
+      "step": 404
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9224137976751797e-05,
+      "loss": 0.7398,
+      "step": 405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921941929689715e-05,
+      "loss": 0.7904,
+      "step": 406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9214686894051088e-05,
+      "loss": 0.7799,
+      "step": 407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920994077525773e-05,
+      "loss": 0.7874,
+      "step": 408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920518094758161e-05,
+      "loss": 0.7272,
+      "step": 409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920040741810768e-05,
+      "loss": 0.7362,
+      "step": 410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919562019394128e-05,
+      "loss": 0.7558,
+      "step": 411
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9190819282208123e-05,
+      "loss": 0.7404,
+      "step": 412
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9186004690054316e-05,
+      "loss": 0.7378,
+      "step": 413
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918117642464632e-05,
+      "loss": 0.7029,
+      "step": 414
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917633449317095e-05,
+      "loss": 0.6837,
+      "step": 415
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917147890283535e-05,
+      "loss": 0.7883,
+      "step": 416
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9166609660867016e-05,
+      "loss": 0.7353,
+      "step": 417
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9161726774513748e-05,
+      "loss": 0.7311,
+      "step": 418
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915683025104367e-05,
+      "loss": 0.7815,
+      "step": 419
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915192009774519e-05,
+      "loss": 0.8526,
+      "step": 420
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9146996321927012e-05,
+      "loss": 0.6514,
+      "step": 421
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9142058930918116e-05,
+      "loss": 0.7148,
+      "step": 422
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9137107932067746e-05,
+      "loss": 0.8271,
+      "step": 423
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9132143332745404e-05,
+      "loss": 0.7014,
+      "step": 424
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9127165140340837e-05,
+      "loss": 0.7621,
+      "step": 425
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9122173362264015e-05,
+      "loss": 0.7356,
+      "step": 426
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9117168005945144e-05,
+      "loss": 0.7188,
+      "step": 427
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9112149078834634e-05,
+      "loss": 0.7882,
+      "step": 428
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9107116588403098e-05,
+      "loss": 0.7465,
+      "step": 429
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910207054214133e-05,
+      "loss": 0.6798,
+      "step": 430
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909701094756031e-05,
+      "loss": 0.7943,
+      "step": 431
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9091937812191183e-05,
+      "loss": 0.6323,
+      "step": 432
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9086851143585242e-05,
+      "loss": 0.7241,
+      "step": 433
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9081750949313945e-05,
+      "loss": 0.7065,
+      "step": 434
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9076637236968852e-05,
+      "loss": 0.7455,
+      "step": 435
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907151001416167e-05,
+      "loss": 0.7266,
+      "step": 436
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.90663692885242e-05,
+      "loss": 0.6636,
+      "step": 437
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9061215067708355e-05,
+      "loss": 0.7529,
+      "step": 438
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9056047359386122e-05,
+      "loss": 0.7875,
+      "step": 439
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905086617124958e-05,
+      "loss": 0.6999,
+      "step": 440
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9045671511010854e-05,
+      "loss": 0.7326,
+      "step": 441
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9040463386402137e-05,
+      "loss": 0.611,
+      "step": 442
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9035241805175655e-05,
+      "loss": 0.722,
+      "step": 443
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903000677510367e-05,
+      "loss": 0.7585,
+      "step": 444
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9024758303978457e-05,
+      "loss": 0.7977,
+      "step": 445
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.90194963996123e-05,
+      "loss": 0.7443,
+      "step": 446
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9014221069837483e-05,
+      "loss": 0.7317,
+      "step": 447
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9008932322506264e-05,
+      "loss": 0.7216,
+      "step": 448
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900363016549088e-05,
+      "loss": 0.7359,
+      "step": 449
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8998314606683522e-05,
+      "loss": 0.7267,
+      "step": 450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8992985653996342e-05,
+      "loss": 0.726,
+      "step": 451
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8987643315361414e-05,
+      "loss": 0.6849,
+      "step": 452
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898228759873074e-05,
+      "loss": 0.7388,
+      "step": 453
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8976918512076245e-05,
+      "loss": 0.761,
+      "step": 454
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8971536063389745e-05,
+      "loss": 0.7297,
+      "step": 455
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8966140260682945e-05,
+      "loss": 0.6795,
+      "step": 456
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8960731111987437e-05,
+      "loss": 0.8232,
+      "step": 457
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8955308625354664e-05,
+      "loss": 0.7866,
+      "step": 458
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8949872808855935e-05,
+      "loss": 0.6882,
+      "step": 459
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.89444236705824e-05,
+      "loss": 0.7191,
+      "step": 460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8938961218645024e-05,
+      "loss": 0.7098,
+      "step": 461
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8933485461174603e-05,
+      "loss": 0.6992,
+      "step": 462
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8927996406321737e-05,
+      "loss": 0.6704,
+      "step": 463
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8922494062256817e-05,
+      "loss": 0.6545,
+      "step": 464
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8916978437170006e-05,
+      "loss": 0.7373,
+      "step": 465
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891144953927125e-05,
+      "loss": 0.7537,
+      "step": 466
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8905907376790247e-05,
+      "loss": 0.7278,
+      "step": 467
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8900351957976434e-05,
+      "loss": 0.7112,
+      "step": 468
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8894783291098977e-05,
+      "loss": 0.689,
+      "step": 469
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888920138444678e-05,
+      "loss": 0.7157,
+      "step": 470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888360624632843e-05,
+      "loss": 0.6744,
+      "step": 471
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887799788507223e-05,
+      "loss": 0.7203,
+      "step": 472
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887237630902615e-05,
+      "loss": 0.6783,
+      "step": 473
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8866741526557844e-05,
+      "loss": 0.6766,
+      "step": 474
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8861093546054605e-05,
+      "loss": 0.7837,
+      "step": 475
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885543237592339e-05,
+      "loss": 0.689,
+      "step": 476
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884975802459078e-05,
+      "loss": 0.6974,
+      "step": 477
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8844070500502972e-05,
+      "loss": 0.8179,
+      "step": 478
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8838369812125782e-05,
+      "loss": 0.7655,
+      "step": 479
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8832655967944607e-05,
+      "loss": 0.7701,
+      "step": 480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882692897646444e-05,
+      "loss": 0.6889,
+      "step": 481
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8821188846209834e-05,
+      "loss": 0.7135,
+      "step": 482
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8815435585724898e-05,
+      "loss": 0.7147,
+      "step": 483
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8809669203573292e-05,
+      "loss": 0.6748,
+      "step": 484
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8803889708338205e-05,
+      "loss": 0.7159,
+      "step": 485
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8798097108622347e-05,
+      "loss": 0.6742,
+      "step": 486
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8792291413047924e-05,
+      "loss": 0.7795,
+      "step": 487
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8786472630256647e-05,
+      "loss": 0.644,
+      "step": 488
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.87806407689097e-05,
+      "loss": 0.6917,
+      "step": 489
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877479583768774e-05,
+      "loss": 0.7315,
+      "step": 490
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8768937845290865e-05,
+      "loss": 0.6957,
+      "step": 491
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8763066800438638e-05,
+      "loss": 0.7071,
+      "step": 492
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8757182711870028e-05,
+      "loss": 0.7199,
+      "step": 493
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8751285588343427e-05,
+      "loss": 0.7821,
+      "step": 494
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8745375438636632e-05,
+      "loss": 0.7498,
+      "step": 495
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873945227154683e-05,
+      "loss": 0.7491,
+      "step": 496
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873351609589058e-05,
+      "loss": 0.7743,
+      "step": 497
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8727566920503806e-05,
+      "loss": 0.7521,
+      "step": 498
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8721604754241784e-05,
+      "loss": 0.7405,
+      "step": 499
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.871562960597912e-05,
+      "loss": 0.7231,
+      "step": 500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870964148460975e-05,
+      "loss": 0.6913,
+      "step": 501
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870364039904692e-05,
+      "loss": 0.7771,
+      "step": 502
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8697626358223172e-05,
+      "loss": 0.7087,
+      "step": 503
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869159937109033e-05,
+      "loss": 0.7702,
+      "step": 504
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868555944661949e-05,
+      "loss": 0.6576,
+      "step": 505
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8679506593801005e-05,
+      "loss": 0.6692,
+      "step": 506
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867344082164447e-05,
+      "loss": 0.8231,
+      "step": 507
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866736213917871e-05,
+      "loss": 0.7598,
+      "step": 508
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8661270555451774e-05,
+      "loss": 0.6594,
+      "step": 509
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8655166079530906e-05,
+      "loss": 0.6528,
+      "step": 510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.864904872050254e-05,
+      "loss": 0.6745,
+      "step": 511
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8642918487472296e-05,
+      "loss": 0.7047,
+      "step": 512
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8636775389564943e-05,
+      "loss": 0.7526,
+      "step": 513
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863061943592441e-05,
+      "loss": 0.7252,
+      "step": 514
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862445063571376e-05,
+      "loss": 0.7448,
+      "step": 515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8618268998115168e-05,
+      "loss": 0.6607,
+      "step": 516
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8612074532329934e-05,
+      "loss": 0.6808,
+      "step": 517
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8605867247578434e-05,
+      "loss": 0.7326,
+      "step": 518
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859964715310014e-05,
+      "loss": 0.7778,
+      "step": 519
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8593414258153588e-05,
+      "loss": 0.7209,
+      "step": 520
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8587168572016362e-05,
+      "loss": 0.706,
+      "step": 521
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858091010398509e-05,
+      "loss": 0.7294,
+      "step": 522
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857463886337542e-05,
+      "loss": 0.7084,
+      "step": 523
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856835485952202e-05,
+      "loss": 0.7257,
+      "step": 524
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856205810177855e-05,
+      "loss": 0.7191,
+      "step": 525
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8555748599517652e-05,
+      "loss": 0.6769,
+      "step": 526
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8549426362130952e-05,
+      "loss": 0.6707,
+      "step": 527
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8543091399029013e-05,
+      "loss": 0.6689,
+      "step": 528
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8536743719641356e-05,
+      "loss": 0.7914,
+      "step": 529
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853038333341642e-05,
+      "loss": 0.7441,
+      "step": 530
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8524010249821554e-05,
+      "loss": 0.6718,
+      "step": 531
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8517624478343025e-05,
+      "loss": 0.6975,
+      "step": 532
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8511226028485973e-05,
+      "loss": 0.6916,
+      "step": 533
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8504814909774407e-05,
+      "loss": 0.6856,
+      "step": 534
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.84983911317512e-05,
+      "loss": 0.7188,
+      "step": 535
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8491954703978067e-05,
+      "loss": 0.7307,
+      "step": 536
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8485505636035544e-05,
+      "loss": 0.6792,
+      "step": 537
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8479043937522996e-05,
+      "loss": 0.6748,
+      "step": 538
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8472569618058585e-05,
+      "loss": 0.7302,
+      "step": 539
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8466082687279247e-05,
+      "loss": 0.6579,
+      "step": 540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8459583154840698e-05,
+      "loss": 0.6318,
+      "step": 541
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845307103041742e-05,
+      "loss": 0.7378,
+      "step": 542
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844654632370262e-05,
+      "loss": 0.7645,
+      "step": 543
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8440009044408248e-05,
+      "loss": 0.7594,
+      "step": 544
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8433459202264963e-05,
+      "loss": 0.8069,
+      "step": 545
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842689680702212e-05,
+      "loss": 0.6601,
+      "step": 546
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842032186844777e-05,
+      "loss": 0.7652,
+      "step": 547
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8413734396328626e-05,
+      "loss": 0.7404,
+      "step": 548
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8407134400470054e-05,
+      "loss": 0.7207,
+      "step": 549
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8400521890696068e-05,
+      "loss": 0.7515,
+      "step": 550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839389687684931e-05,
+      "loss": 0.7448,
+      "step": 551
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8387259368791027e-05,
+      "loss": 0.7275,
+      "step": 552
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8380609376401072e-05,
+      "loss": 0.7327,
+      "step": 553
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837394690957787e-05,
+      "loss": 0.7036,
+      "step": 554
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8367271978238422e-05,
+      "loss": 0.808,
+      "step": 555
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8360584592318286e-05,
+      "loss": 0.6785,
+      "step": 556
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8353884761771546e-05,
+      "loss": 0.7284,
+      "step": 557
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8347172496570825e-05,
+      "loss": 0.7904,
+      "step": 558
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834044780670724e-05,
+      "loss": 0.6307,
+      "step": 559
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833371070219041e-05,
+      "loss": 0.6645,
+      "step": 560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8326961193048434e-05,
+      "loss": 0.6836,
+      "step": 561
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8320199289327866e-05,
+      "loss": 0.7199,
+      "step": 562
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8313425001093724e-05,
+      "loss": 0.6821,
+      "step": 563
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830663833842945e-05,
+      "loss": 0.7246,
+      "step": 564
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8299839311436905e-05,
+      "loss": 0.679,
+      "step": 565
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829302793023636e-05,
+      "loss": 0.6527,
+      "step": 566
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8286204204966465e-05,
+      "loss": 0.6961,
+      "step": 567
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827936814578426e-05,
+      "loss": 0.6864,
+      "step": 568
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8272519762865123e-05,
+      "loss": 0.6592,
+      "step": 569
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8265659066402794e-05,
+      "loss": 0.7445,
+      "step": 570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.825878606660934e-05,
+      "loss": 0.7062,
+      "step": 571
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8251900773715124e-05,
+      "loss": 0.7635,
+      "step": 572
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824500319796883e-05,
+      "loss": 0.682,
+      "step": 573
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8238093349637404e-05,
+      "loss": 0.7563,
+      "step": 574
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8231171239006077e-05,
+      "loss": 0.7939,
+      "step": 575
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822423687637832e-05,
+      "loss": 0.6267,
+      "step": 576
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8217290272075855e-05,
+      "loss": 0.6508,
+      "step": 577
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8210331436438607e-05,
+      "loss": 0.7355,
+      "step": 578
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8203360379824722e-05,
+      "loss": 0.704,
+      "step": 579
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8196377112610524e-05,
+      "loss": 0.6941,
+      "step": 580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8189381645190533e-05,
+      "loss": 0.6892,
+      "step": 581
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8182373987977404e-05,
+      "loss": 0.7523,
+      "step": 582
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817535415140195e-05,
+      "loss": 0.7297,
+      "step": 583
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816832214591311e-05,
+      "loss": 0.7905,
+      "step": 584
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8161277981977942e-05,
+      "loss": 0.722,
+      "step": 585
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8154221670081595e-05,
+      "loss": 0.6798,
+      "step": 586
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8147153220727302e-05,
+      "loss": 0.7355,
+      "step": 587
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8140072644436357e-05,
+      "loss": 0.7314,
+      "step": 588
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.813297995174812e-05,
+      "loss": 0.7347,
+      "step": 589
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8125875153219966e-05,
+      "loss": 0.7302,
+      "step": 590
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8118758259427305e-05,
+      "loss": 0.6914,
+      "step": 591
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8111629280963546e-05,
+      "loss": 0.6765,
+      "step": 592
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8104488228440083e-05,
+      "loss": 0.7157,
+      "step": 593
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8097335112486287e-05,
+      "loss": 0.7092,
+      "step": 594
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 0.676,
+      "step": 595
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.808299273289492e-05,
+      "loss": 0.7315,
+      "step": 596
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.807580349060581e-05,
+      "loss": 0.6873,
+      "step": 597
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8068602227583242e-05,
+      "loss": 0.7107,
+      "step": 598
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8061388954546202e-05,
+      "loss": 0.6639,
+      "step": 599
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8054163682231567e-05,
+      "loss": 0.6641,
+      "step": 600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804692642139406e-05,
+      "loss": 0.691,
+      "step": 601
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803967718280625e-05,
+      "loss": 0.7241,
+      "step": 602
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8032415977258552e-05,
+      "loss": 0.7186,
+      "step": 603
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8025142815559164e-05,
+      "loss": 0.7607,
+      "step": 604
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8017857708534107e-05,
+      "loss": 0.74,
+      "step": 605
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8010560667027174e-05,
+      "loss": 0.6744,
+      "step": 606
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8003251701899913e-05,
+      "loss": 0.7783,
+      "step": 607
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7995930824031632e-05,
+      "loss": 0.7881,
+      "step": 608
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7988598044319373e-05,
+      "loss": 0.7612,
+      "step": 609
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798125337367788e-05,
+      "loss": 0.6694,
+      "step": 610
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7973896823039604e-05,
+      "loss": 0.7516,
+      "step": 611
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7966528403354685e-05,
+      "loss": 0.6845,
+      "step": 612
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795914812559092e-05,
+      "loss": 0.7317,
+      "step": 613
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7951756000733764e-05,
+      "loss": 0.7508,
+      "step": 614
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.79443520397863e-05,
+      "loss": 0.7118,
+      "step": 615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7936936253769234e-05,
+      "loss": 0.7209,
+      "step": 616
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7929508653720872e-05,
+      "loss": 0.6808,
+      "step": 617
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7922069250697105e-05,
+      "loss": 0.7119,
+      "step": 618
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7914618055771392e-05,
+      "loss": 0.7175,
+      "step": 619
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790715508003474e-05,
+      "loss": 0.7466,
+      "step": 620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7899680334595703e-05,
+      "loss": 0.7571,
+      "step": 621
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7892193830580344e-05,
+      "loss": 0.7657,
+      "step": 622
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7884695579132233e-05,
+      "loss": 0.7996,
+      "step": 623
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7877185591412418e-05,
+      "loss": 0.7698,
+      "step": 624
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.786966387859943e-05,
+      "loss": 0.6654,
+      "step": 625
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7862130451889243e-05,
+      "loss": 0.7077,
+      "step": 626
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785458532249527e-05,
+      "loss": 0.776,
+      "step": 627
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784702850164834e-05,
+      "loss": 0.735,
+      "step": 628
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783946000059669e-05,
+      "loss": 0.7477,
+      "step": 629
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783187983060594e-05,
+      "loss": 0.6533,
+      "step": 630
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7824288002959074e-05,
+      "loss": 0.7409,
+      "step": 631
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7816684528956437e-05,
+      "loss": 0.7719,
+      "step": 632
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.780906941991571e-05,
+      "loss": 0.6925,
+      "step": 633
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7801442687171887e-05,
+      "loss": 0.7015,
+      "step": 634
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7793804342077258e-05,
+      "loss": 0.7051,
+      "step": 635
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778615439600141e-05,
+      "loss": 0.6838,
+      "step": 636
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.77784928603312e-05,
+      "loss": 0.6936,
+      "step": 637
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7770819746470717e-05,
+      "loss": 0.7746,
+      "step": 638
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.77631350658413e-05,
+      "loss": 0.6731,
+      "step": 639
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7755438829881503e-05,
+      "loss": 0.6915,
+      "step": 640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7747731050047078e-05,
+      "loss": 0.6168,
+      "step": 641
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7740011737810955e-05,
+      "loss": 0.694,
+      "step": 642
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773228090466324e-05,
+      "loss": 0.6969,
+      "step": 643
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7724538562111183e-05,
+      "loss": 0.7154,
+      "step": 644
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771678472167916e-05,
+      "loss": 0.7546,
+      "step": 645
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7709019394908672e-05,
+      "loss": 0.7072,
+      "step": 646
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7701242593358307e-05,
+      "loss": 0.7471,
+      "step": 647
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769345432860374e-05,
+      "loss": 0.6253,
+      "step": 648
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7685654612237705e-05,
+      "loss": 0.6968,
+      "step": 649
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7677843455869984e-05,
+      "loss": 0.6984,
+      "step": 650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7670020871127386e-05,
+      "loss": 0.6286,
+      "step": 651
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7662186869653735e-05,
+      "loss": 0.7112,
+      "step": 652
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765434146310984e-05,
+      "loss": 0.7195,
+      "step": 653
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7646484663173496e-05,
+      "loss": 0.7288,
+      "step": 654
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763861648153945e-05,
+      "loss": 0.6833,
+      "step": 655
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7630736929919396e-05,
+      "loss": 0.7173,
+      "step": 656
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.762284602004194e-05,
+      "loss": 0.6912,
+      "step": 657
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7614943763652614e-05,
+      "loss": 0.64,
+      "step": 658
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7607030172513823e-05,
+      "loss": 0.722,
+      "step": 659
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759910525840485e-05,
+      "loss": 0.6999,
+      "step": 660
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759116903312183e-05,
+      "loss": 0.6614,
+      "step": 661
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7583221508477746e-05,
+      "loss": 0.7665,
+      "step": 662
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7575262696302378e-05,
+      "loss": 0.6969,
+      "step": 663
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756729260844232e-05,
+      "loss": 0.7083,
+      "step": 664
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7559311256760958e-05,
+      "loss": 0.6955,
+      "step": 665
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755131865313843e-05,
+      "loss": 0.6959,
+      "step": 666
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7543314809471624e-05,
+      "loss": 0.629,
+      "step": 667
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753529973767417e-05,
+      "loss": 0.6775,
+      "step": 668
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7527273449676395e-05,
+      "loss": 0.7409,
+      "step": 669
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7519235957425334e-05,
+      "loss": 0.7177,
+      "step": 670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7511187272884695e-05,
+      "loss": 0.7264,
+      "step": 671
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750312740803484e-05,
+      "loss": 0.7196,
+      "step": 672
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7495056374872785e-05,
+      "loss": 0.6876,
+      "step": 673
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7486974185412157e-05,
+      "loss": 0.7016,
+      "step": 674
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.74788808516832e-05,
+      "loss": 0.6493,
+      "step": 675
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747077638573274e-05,
+      "loss": 0.7015,
+      "step": 676
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7462660799624166e-05,
+      "loss": 0.6966,
+      "step": 677
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7454534105437438e-05,
+      "loss": 0.6743,
+      "step": 678
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.744639631526904e-05,
+      "loss": 0.6614,
+      "step": 679
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743824744123196e-05,
+      "loss": 0.694,
+      "step": 680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743008749545571e-05,
+      "loss": 0.6486,
+      "step": 681
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742191649008626e-05,
+      "loss": 0.7054,
+      "step": 682
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741373443728605e-05,
+      "loss": 0.6772,
+      "step": 683
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7405541349233968e-05,
+      "loss": 0.7208,
+      "step": 684
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.739733723812532e-05,
+      "loss": 0.7527,
+      "step": 685
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7389122116171832e-05,
+      "loss": 0.6728,
+      "step": 686
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.73808959956016e-05,
+      "loss": 0.7563,
+      "step": 687
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.737265888865911e-05,
+      "loss": 0.7535,
+      "step": 688
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7364410807605193e-05,
+      "loss": 0.6854,
+      "step": 689
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7356151764717012e-05,
+      "loss": 0.7318,
+      "step": 690
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.734788177228805e-05,
+      "loss": 0.6674,
+      "step": 691
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7339600842628094e-05,
+      "loss": 0.7021,
+      "step": 692
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.73313089880632e-05,
+      "loss": 0.7252,
+      "step": 693
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7323006220935696e-05,
+      "loss": 0.7269,
+      "step": 694
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7314692553604143e-05,
+      "loss": 0.6243,
+      "step": 695
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7306367998443333e-05,
+      "loss": 0.7536,
+      "step": 696
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7298032567844264e-05,
+      "loss": 0.6759,
+      "step": 697
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7289686274214116e-05,
+      "loss": 0.6683,
+      "step": 698
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7281329129976253e-05,
+      "loss": 0.7551,
+      "step": 699
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7272961147570177e-05,
+      "loss": 0.7218,
+      "step": 700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7264582339451525e-05,
+      "loss": 0.7232,
+      "step": 701
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7256192718092046e-05,
+      "loss": 0.6767,
+      "step": 702
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7247792295979593e-05,
+      "loss": 0.6837,
+      "step": 703
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7239381085618085e-05,
+      "loss": 0.7191,
+      "step": 704
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7230959099527512e-05,
+      "loss": 0.6622,
+      "step": 705
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7222526350243893e-05,
+      "loss": 0.6633,
+      "step": 706
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7214082850319265e-05,
+      "loss": 0.6663,
+      "step": 707
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720562861232168e-05,
+      "loss": 0.6753,
+      "step": 708
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7197163648835163e-05,
+      "loss": 0.6539,
+      "step": 709
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7188687972459707e-05,
+      "loss": 0.6867,
+      "step": 710
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7180201595811254e-05,
+      "loss": 0.6714,
+      "step": 711
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717170453152167e-05,
+      "loss": 0.6438,
+      "step": 712
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.716319679223873e-05,
+      "loss": 0.6976,
+      "step": 713
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7154678390626093e-05,
+      "loss": 0.709,
+      "step": 714
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.71461493393633e-05,
+      "loss": 0.6678,
+      "step": 715
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7137609651145738e-05,
+      "loss": 0.7199,
+      "step": 716
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7129059338684626e-05,
+      "loss": 0.7096,
+      "step": 717
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7120498414707e-05,
+      "loss": 0.6812,
+      "step": 718
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7111926891955684e-05,
+      "loss": 0.7046,
+      "step": 719
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7103344783189292e-05,
+      "loss": 0.7106,
+      "step": 720
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7094752101182175e-05,
+      "loss": 0.7035,
+      "step": 721
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7086148858724445e-05,
+      "loss": 0.6803,
+      "step": 722
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7077535068621916e-05,
+      "loss": 0.7451,
+      "step": 723
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706891074369611e-05,
+      "loss": 0.6783,
+      "step": 724
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7060275896784225e-05,
+      "loss": 0.7182,
+      "step": 725
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705163054073913e-05,
+      "loss": 0.6527,
+      "step": 726
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704297468842932e-05,
+      "loss": 0.7512,
+      "step": 727
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703430835273893e-05,
+      "loss": 0.7429,
+      "step": 728
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7025631546567698e-05,
+      "loss": 0.647,
+      "step": 729
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7016944282830935e-05,
+      "loss": 0.6808,
+      "step": 730
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7008246574459527e-05,
+      "loss": 0.6505,
+      "step": 731
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6999538434399908e-05,
+      "loss": 0.7316,
+      "step": 732
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6990819875614033e-05,
+      "loss": 0.7029,
+      "step": 733
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6982090911079372e-05,
+      "loss": 0.7339,
+      "step": 734
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.697335155378888e-05,
+      "loss": 0.7925,
+      "step": 735
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6964601816750977e-05,
+      "loss": 0.7677,
+      "step": 736
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695584171298955e-05,
+      "loss": 0.7428,
+      "step": 737
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6947071255543894e-05,
+      "loss": 0.7016,
+      "step": 738
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6938290457468728e-05,
+      "loss": 0.7199,
+      "step": 739
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692949933183416e-05,
+      "loss": 0.6714,
+      "step": 740
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692069789172568e-05,
+      "loss": 0.6802,
+      "step": 741
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6911886150244114e-05,
+      "loss": 0.6907,
+      "step": 742
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6903064120505638e-05,
+      "loss": 0.7091,
+      "step": 743
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6894231815641723e-05,
+      "loss": 0.7145,
+      "step": 744
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6885389248799153e-05,
+      "loss": 0.6579,
+      "step": 745
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687653643313998e-05,
+      "loss": 0.6616,
+      "step": 746
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6867673381841508e-05,
+      "loss": 0.7192,
+      "step": 747
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6858800108096277e-05,
+      "loss": 0.6716,
+      "step": 748
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.684991662511205e-05,
+      "loss": 0.6539,
+      "step": 749
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6841022946111774e-05,
+      "loss": 0.6842,
+      "step": 750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6832119084333588e-05,
+      "loss": 0.6516,
+      "step": 751
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6823205053030777e-05,
+      "loss": 0.7063,
+      "step": 752
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681428086547176e-05,
+      "loss": 0.6887,
+      "step": 753
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6805346534940096e-05,
+      "loss": 0.7392,
+      "step": 754
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6796402074734404e-05,
+      "loss": 0.6477,
+      "step": 755
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6787447498168418e-05,
+      "loss": 0.7135,
+      "step": 756
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6778482818570912e-05,
+      "loss": 0.7042,
+      "step": 757
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.676950804928569e-05,
+      "loss": 0.656,
+      "step": 758
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.67605232036716e-05,
+      "loss": 0.7285,
+      "step": 759
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.675152829510246e-05,
+      "loss": 0.7185,
+      "step": 760
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.674252333696709e-05,
+      "loss": 0.651,
+      "step": 761
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6733508342669256e-05,
+      "loss": 0.6685,
+      "step": 762
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672448332562766e-05,
+      "loss": 0.6821,
+      "step": 763
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6715448299275935e-05,
+      "loss": 0.7201,
+      "step": 764
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.67064032770626e-05,
+      "loss": 0.6725,
+      "step": 765
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6697348272451066e-05,
+      "loss": 0.6833,
+      "step": 766
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6688283298919588e-05,
+      "loss": 0.7563,
+      "step": 767
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.667920836996127e-05,
+      "loss": 0.6257,
+      "step": 768
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.667012349908403e-05,
+      "loss": 0.7686,
+      "step": 769
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666102869981059e-05,
+      "loss": 0.6859,
+      "step": 770
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6651923985678436e-05,
+      "loss": 0.6642,
+      "step": 771
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6642809370239833e-05,
+      "loss": 0.7175,
+      "step": 772
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.663368486706177e-05,
+      "loss": 0.683,
+      "step": 773
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6624550489725952e-05,
+      "loss": 0.6961,
+      "step": 774
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6615406251828794e-05,
+      "loss": 0.6556,
+      "step": 775
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6606252166981378e-05,
+      "loss": 0.6738,
+      "step": 776
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6597088248809438e-05,
+      "loss": 0.6604,
+      "step": 777
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6587914510953366e-05,
+      "loss": 0.6649,
+      "step": 778
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657873096706815e-05,
+      "loss": 0.7124,
+      "step": 779
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6569537630823385e-05,
+      "loss": 0.6872,
+      "step": 780
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6560334515903232e-05,
+      "loss": 0.7371,
+      "step": 781
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6551121636006414e-05,
+      "loss": 0.6671,
+      "step": 782
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.65418990048462e-05,
+      "loss": 0.7499,
+      "step": 783
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6532666636150348e-05,
+      "loss": 0.698,
+      "step": 784
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.652342454366113e-05,
+      "loss": 0.6667,
+      "step": 785
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6514172741135288e-05,
+      "loss": 0.727,
+      "step": 786
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6504911242344016e-05,
+      "loss": 0.6331,
+      "step": 787
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6495640061072933e-05,
+      "loss": 0.6606,
+      "step": 788
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6486359211122088e-05,
+      "loss": 0.6468,
+      "step": 789
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.64770687063059e-05,
+      "loss": 0.6669,
+      "step": 790
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6467768560453184e-05,
+      "loss": 0.7135,
+      "step": 791
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.645845878740708e-05,
+      "loss": 0.7076,
+      "step": 792
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.644913940102507e-05,
+      "loss": 0.7262,
+      "step": 793
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6439810415178958e-05,
+      "loss": 0.6494,
+      "step": 794
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6430471843754806e-05,
+      "loss": 0.7307,
+      "step": 795
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6421123700652973e-05,
+      "loss": 0.7144,
+      "step": 796
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6411765999788057e-05,
+      "loss": 0.6608,
+      "step": 797
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640239875508887e-05,
+      "loss": 0.7182,
+      "step": 798
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6393021980498446e-05,
+      "loss": 0.6755,
+      "step": 799
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6383635689973997e-05,
+      "loss": 0.6836,
+      "step": 800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.63742398974869e-05,
+      "loss": 0.6592,
+      "step": 801
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6364834617022674e-05,
+      "loss": 0.606,
+      "step": 802
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6355419862580963e-05,
+      "loss": 0.6596,
+      "step": 803
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6345995648175516e-05,
+      "loss": 0.7467,
+      "step": 804
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6336561987834155e-05,
+      "loss": 0.6506,
+      "step": 805
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6327118895598765e-05,
+      "loss": 0.6279,
+      "step": 806
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631766638552528e-05,
+      "loss": 0.6743,
+      "step": 807
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6308204471683638e-05,
+      "loss": 0.6912,
+      "step": 808
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.629873316815778e-05,
+      "loss": 0.7616,
+      "step": 809
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6289252489045625e-05,
+      "loss": 0.7234,
+      "step": 810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.627976244845905e-05,
+      "loss": 0.6367,
+      "step": 811
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6270263060523856e-05,
+      "loss": 0.6891,
+      "step": 812
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626075433937977e-05,
+      "loss": 0.7373,
+      "step": 813
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6251236299180405e-05,
+      "loss": 0.6238,
+      "step": 814
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6241708954093242e-05,
+      "loss": 0.6845,
+      "step": 815
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.623217231829962e-05,
+      "loss": 0.7082,
+      "step": 816
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6222626405994703e-05,
+      "loss": 0.7489,
+      "step": 817
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6213071231387463e-05,
+      "loss": 0.752,
+      "step": 818
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6203506808700656e-05,
+      "loss": 0.7575,
+      "step": 819
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6193933152170812e-05,
+      "loss": 0.7331,
+      "step": 820
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6184350276048194e-05,
+      "loss": 0.6567,
+      "step": 821
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.61747581945968e-05,
+      "loss": 0.716,
+      "step": 822
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.616515692209432e-05,
+      "loss": 0.673,
+      "step": 823
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6155546472832132e-05,
+      "loss": 0.6659,
+      "step": 824
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.614592686111527e-05,
+      "loss": 0.721,
+      "step": 825
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6136298101262407e-05,
+      "loss": 0.681,
+      "step": 826
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6126660207605834e-05,
+      "loss": 0.6567,
+      "step": 827
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6117013194491434e-05,
+      "loss": 0.7218,
+      "step": 828
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6107357076278664e-05,
+      "loss": 0.6574,
+      "step": 829
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6097691867340547e-05,
+      "loss": 0.6984,
+      "step": 830
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6088017582063612e-05,
+      "loss": 0.7142,
+      "step": 831
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6078334234847923e-05,
+      "loss": 0.7592,
+      "step": 832
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.606864184010702e-05,
+      "loss": 0.6532,
+      "step": 833
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6058940412267912e-05,
+      "loss": 0.7941,
+      "step": 834
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6049229965771054e-05,
+      "loss": 0.622,
+      "step": 835
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6039510515070325e-05,
+      "loss": 0.6774,
+      "step": 836
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6029782074633013e-05,
+      "loss": 0.6319,
+      "step": 837
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6020044658939767e-05,
+      "loss": 0.7015,
+      "step": 838
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6010298282484623e-05,
+      "loss": 0.7048,
+      "step": 839
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.600054295977494e-05,
+      "loss": 0.6395,
+      "step": 840
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5990778705331385e-05,
+      "loss": 0.7322,
+      "step": 841
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.598100553368794e-05,
+      "loss": 0.6668,
+      "step": 842
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5971223459391853e-05,
+      "loss": 0.6981,
+      "step": 843
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5961432497003606e-05,
+      "loss": 0.757,
+      "step": 844
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5951632661096932e-05,
+      "loss": 0.651,
+      "step": 845
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5941823966258774e-05,
+      "loss": 0.6855,
+      "step": 846
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593200642708924e-05,
+      "loss": 0.6252,
+      "step": 847
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5922180058201623e-05,
+      "loss": 0.6248,
+      "step": 848
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5912344874222353e-05,
+      "loss": 0.6804,
+      "step": 849
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.590250088979097e-05,
+      "loss": 0.7239,
+      "step": 850
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.589264811956013e-05,
+      "loss": 0.6088,
+      "step": 851
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5882786578195557e-05,
+      "loss": 0.6383,
+      "step": 852
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.587291628037604e-05,
+      "loss": 0.6902,
+      "step": 853
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5863037240793383e-05,
+      "loss": 0.7232,
+      "step": 854
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.585314947415242e-05,
+      "loss": 0.6416,
+      "step": 855
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5843252995170973e-05,
+      "loss": 0.6992,
+      "step": 856
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5833347818579825e-05,
+      "loss": 0.6597,
+      "step": 857
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.582343395912271e-05,
+      "loss": 0.6994,
+      "step": 858
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.581351143155628e-05,
+      "loss": 0.6412,
+      "step": 859
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5803580250650098e-05,
+      "loss": 0.7192,
+      "step": 860
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5793640431186604e-05,
+      "loss": 0.6786,
+      "step": 861
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5783691987961098e-05,
+      "loss": 0.6478,
+      "step": 862
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.577373493578171e-05,
+      "loss": 0.6302,
+      "step": 863
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.576376928946939e-05,
+      "loss": 0.6305,
+      "step": 864
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5753795063857886e-05,
+      "loss": 0.6726,
+      "step": 865
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57438122737937e-05,
+      "loss": 0.626,
+      "step": 866
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5733820934136096e-05,
+      "loss": 0.6387,
+      "step": 867
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5723821059757057e-05,
+      "loss": 0.6869,
+      "step": 868
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5713812665541273e-05,
+      "loss": 0.7504,
+      "step": 869
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5703795766386114e-05,
+      "loss": 0.7439,
+      "step": 870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5693770377201613e-05,
+      "loss": 0.6655,
+      "step": 871
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5683736512910436e-05,
+      "loss": 0.7097,
+      "step": 872
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5673694188447865e-05,
+      "loss": 0.6837,
+      "step": 873
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5663643418761777e-05,
+      "loss": 0.6555,
+      "step": 874
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565358421881262e-05,
+      "loss": 0.6274,
+      "step": 875
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.564351660357338e-05,
+      "loss": 0.735,
+      "step": 876
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.563344058802959e-05,
+      "loss": 0.6409,
+      "step": 877
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5623356187179265e-05,
+      "loss": 0.6634,
+      "step": 878
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5613263416032914e-05,
+      "loss": 0.6605,
+      "step": 879
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5603162289613503e-05,
+      "loss": 0.7145,
+      "step": 880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5593052822956426e-05,
+      "loss": 0.7608,
+      "step": 881
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5582935031109513e-05,
+      "loss": 0.6883,
+      "step": 882
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.557280892913296e-05,
+      "loss": 0.6855,
+      "step": 883
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5562674532099346e-05,
+      "loss": 0.6593,
+      "step": 884
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.55525318550936e-05,
+      "loss": 0.6675,
+      "step": 885
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.554238091321297e-05,
+      "loss": 0.5602,
+      "step": 886
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5532221721567006e-05,
+      "loss": 0.5226,
+      "step": 887
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5522054295277534e-05,
+      "loss": 0.5249,
+      "step": 888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.551187864947865e-05,
+      "loss": 0.4971,
+      "step": 889
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5501694799316672e-05,
+      "loss": 0.4589,
+      "step": 890
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549150275995014e-05,
+      "loss": 0.532,
+      "step": 891
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5481302546549774e-05,
+      "loss": 0.5011,
+      "step": 892
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5471094174298464e-05,
+      "loss": 0.4853,
+      "step": 893
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5460877658391253e-05,
+      "loss": 0.4818,
+      "step": 894
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5450653014035288e-05,
+      "loss": 0.4769,
+      "step": 895
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544042025644983e-05,
+      "loss": 0.4411,
+      "step": 896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5430179400866216e-05,
+      "loss": 0.4618,
+      "step": 897
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5419930462527823e-05,
+      "loss": 0.435,
+      "step": 898
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.540967345669008e-05,
+      "loss": 0.5316,
+      "step": 899
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.539940839862041e-05,
+      "loss": 0.5382,
+      "step": 900
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.538913530359822e-05,
+      "loss": 0.4752,
+      "step": 901
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5378854186914886e-05,
+      "loss": 0.4789,
+      "step": 902
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5368565063873723e-05,
+      "loss": 0.4741,
+      "step": 903
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5358267949789968e-05,
+      "loss": 0.4465,
+      "step": 904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5347962859990744e-05,
+      "loss": 0.4323,
+      "step": 905
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.533764980981505e-05,
+      "loss": 0.4747,
+      "step": 906
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.532732881461373e-05,
+      "loss": 0.4574,
+      "step": 907
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5316999889749466e-05,
+      "loss": 0.4847,
+      "step": 908
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5306663050596723e-05,
+      "loss": 0.3949,
+      "step": 909
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5296318312541768e-05,
+      "loss": 0.4703,
+      "step": 910
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5285965690982614e-05,
+      "loss": 0.4807,
+      "step": 911
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.527560520132901e-05,
+      "loss": 0.4239,
+      "step": 912
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5265236859002406e-05,
+      "loss": 0.4748,
+      "step": 913
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5254860679435965e-05,
+      "loss": 0.5419,
+      "step": 914
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5244476678074496e-05,
+      "loss": 0.4614,
+      "step": 915
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5234084870374455e-05,
+      "loss": 0.4112,
+      "step": 916
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5223685271803923e-05,
+      "loss": 0.4456,
+      "step": 917
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.521327789784257e-05,
+      "loss": 0.4477,
+      "step": 918
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5202862763981645e-05,
+      "loss": 0.5094,
+      "step": 919
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5192439885723942e-05,
+      "loss": 0.4401,
+      "step": 920
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5182009278583794e-05,
+      "loss": 0.4625,
+      "step": 921
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5171570958087025e-05,
+      "loss": 0.5119,
+      "step": 922
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5161124939770946e-05,
+      "loss": 0.5,
+      "step": 923
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5150671239184326e-05,
+      "loss": 0.5293,
+      "step": 924
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.514020987188737e-05,
+      "loss": 0.4577,
+      "step": 925
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5129740853451695e-05,
+      "loss": 0.4637,
+      "step": 926
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5119264199460302e-05,
+      "loss": 0.4643,
+      "step": 927
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5108779925507562e-05,
+      "loss": 0.4623,
+      "step": 928
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5098288047199186e-05,
+      "loss": 0.4557,
+      "step": 929
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5087788580152207e-05,
+      "loss": 0.4705,
+      "step": 930
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.507728153999495e-05,
+      "loss": 0.4291,
+      "step": 931
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5066766942367013e-05,
+      "loss": 0.4664,
+      "step": 932
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5056244802919251e-05,
+      "loss": 0.5382,
+      "step": 933
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.504571513731373e-05,
+      "loss": 0.4752,
+      "step": 934
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5035177961223727e-05,
+      "loss": 0.493,
+      "step": 935
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5024633290333703e-05,
+      "loss": 0.4793,
+      "step": 936
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5014081140339269e-05,
+      "loss": 0.4328,
+      "step": 937
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.500352152694717e-05,
+      "loss": 0.5047,
+      "step": 938
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4992954465875258e-05,
+      "loss": 0.4963,
+      "step": 939
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4982379972852471e-05,
+      "loss": 0.4531,
+      "step": 940
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4971798063618812e-05,
+      "loss": 0.4352,
+      "step": 941
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4961208753925322e-05,
+      "loss": 0.4349,
+      "step": 942
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4950612059534061e-05,
+      "loss": 0.4733,
+      "step": 943
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4940007996218075e-05,
+      "loss": 0.4959,
+      "step": 944
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4929396579761378e-05,
+      "loss": 0.4915,
+      "step": 945
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4918777825958932e-05,
+      "loss": 0.4153,
+      "step": 946
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4908151750616627e-05,
+      "loss": 0.4418,
+      "step": 947
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4897518369551236e-05,
+      "loss": 0.4622,
+      "step": 948
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.488687769859042e-05,
+      "loss": 0.496,
+      "step": 949
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4876229753572688e-05,
+      "loss": 0.431,
+      "step": 950
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4865574550347368e-05,
+      "loss": 0.4649,
+      "step": 951
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4854912104774601e-05,
+      "loss": 0.4962,
+      "step": 952
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4844242432725307e-05,
+      "loss": 0.4315,
+      "step": 953
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4833565550081159e-05,
+      "loss": 0.4822,
+      "step": 954
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4822881472734563e-05,
+      "loss": 0.4635,
+      "step": 955
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.481219021658864e-05,
+      "loss": 0.5026,
+      "step": 956
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4801491797557186e-05,
+      "loss": 0.4592,
+      "step": 957
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4790786231564672e-05,
+      "loss": 0.4994,
+      "step": 958
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4780073534546195e-05,
+      "loss": 0.5324,
+      "step": 959
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4769353722447477e-05,
+      "loss": 0.5082,
+      "step": 960
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4758626811224818e-05,
+      "loss": 0.4462,
+      "step": 961
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4747892816845103e-05,
+      "loss": 0.4274,
+      "step": 962
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.473715175528574e-05,
+      "loss": 0.4551,
+      "step": 963
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.472640364253467e-05,
+      "loss": 0.4719,
+      "step": 964
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4715648494590327e-05,
+      "loss": 0.4722,
+      "step": 965
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4704886327461613e-05,
+      "loss": 0.4437,
+      "step": 966
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4694117157167882e-05,
+      "loss": 0.4727,
+      "step": 967
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4683340999738908e-05,
+      "loss": 0.4392,
+      "step": 968
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4672557871214874e-05,
+      "loss": 0.4449,
+      "step": 969
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4661767787646328e-05,
+      "loss": 0.4687,
+      "step": 970
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4650970765094175e-05,
+      "loss": 0.4436,
+      "step": 971
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4640166819629654e-05,
+      "loss": 0.4411,
+      "step": 972
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4629355967334297e-05,
+      "loss": 0.4432,
+      "step": 973
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.461853822429993e-05,
+      "loss": 0.4778,
+      "step": 974
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4607713606628627e-05,
+      "loss": 0.465,
+      "step": 975
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.45968821304327e-05,
+      "loss": 0.5144,
+      "step": 976
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.458604381183466e-05,
+      "loss": 0.4474,
+      "step": 977
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.457519866696722e-05,
+      "loss": 0.495,
+      "step": 978
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4564346711973237e-05,
+      "loss": 0.4787,
+      "step": 979
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4553487963005712e-05,
+      "loss": 0.469,
+      "step": 980
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4542622436227762e-05,
+      "loss": 0.4426,
+      "step": 981
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4531750147812584e-05,
+      "loss": 0.4131,
+      "step": 982
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4520871113943447e-05,
+      "loss": 0.5014,
+      "step": 983
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.450998535081366e-05,
+      "loss": 0.5042,
+      "step": 984
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4499092874626546e-05,
+      "loss": 0.4907,
+      "step": 985
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4488193701595419e-05,
+      "loss": 0.4584,
+      "step": 986
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4477287847943562e-05,
+      "loss": 0.4521,
+      "step": 987
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4466375329904208e-05,
+      "loss": 0.433,
+      "step": 988
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4455456163720497e-05,
+      "loss": 0.4467,
+      "step": 989
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.444453036564548e-05,
+      "loss": 0.4862,
+      "step": 990
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4433597951942071e-05,
+      "loss": 0.4656,
+      "step": 991
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.442265893888303e-05,
+      "loss": 0.4308,
+      "step": 992
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4411713342750942e-05,
+      "loss": 0.4354,
+      "step": 993
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4400761179838194e-05,
+      "loss": 0.4478,
+      "step": 994
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4389802466446942e-05,
+      "loss": 0.4656,
+      "step": 995
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4378837218889099e-05,
+      "loss": 0.4435,
+      "step": 996
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4367865453486296e-05,
+      "loss": 0.4526,
+      "step": 997
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4356887186569872e-05,
+      "loss": 0.459,
+      "step": 998
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.434590243448084e-05,
+      "loss": 0.4193,
+      "step": 999
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4334911213569872e-05,
+      "loss": 0.4441,
+      "step": 1000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4323913540197257e-05,
+      "loss": 0.4728,
+      "step": 1001
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4312909430732899e-05,
+      "loss": 0.4415,
+      "step": 1002
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4301898901556279e-05,
+      "loss": 0.4821,
+      "step": 1003
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4290881969056436e-05,
+      "loss": 0.4876,
+      "step": 1004
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.427985864963193e-05,
+      "loss": 0.4765,
+      "step": 1005
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4268828959690841e-05,
+      "loss": 0.4757,
+      "step": 1006
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4257792915650728e-05,
+      "loss": 0.4912,
+      "step": 1007
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4246750533938603e-05,
+      "loss": 0.465,
+      "step": 1008
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4235701830990908e-05,
+      "loss": 0.5503,
+      "step": 1009
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4224646823253512e-05,
+      "loss": 0.4634,
+      "step": 1010
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4213585527181647e-05,
+      "loss": 0.4907,
+      "step": 1011
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4202517959239925e-05,
+      "loss": 0.4729,
+      "step": 1012
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4191444135902277e-05,
+      "loss": 0.4742,
+      "step": 1013
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4180364073651953e-05,
+      "loss": 0.4581,
+      "step": 1014
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4169277788981489e-05,
+      "loss": 0.4526,
+      "step": 1015
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4158185298392682e-05,
+      "loss": 0.4723,
+      "step": 1016
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4147086618396566e-05,
+      "loss": 0.5698,
+      "step": 1017
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4135981765513391e-05,
+      "loss": 0.4711,
+      "step": 1018
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4124870756272593e-05,
+      "loss": 0.4698,
+      "step": 1019
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4113753607212768e-05,
+      "loss": 0.4035,
+      "step": 1020
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4102630334881657e-05,
+      "loss": 0.433,
+      "step": 1021
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4091500955836115e-05,
+      "loss": 0.5291,
+      "step": 1022
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4080365486642081e-05,
+      "loss": 0.4808,
+      "step": 1023
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4069223943874567e-05,
+      "loss": 0.4143,
+      "step": 1024
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4058076344117617e-05,
+      "loss": 0.5109,
+      "step": 1025
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4046922703964298e-05,
+      "loss": 0.4565,
+      "step": 1026
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4035763040016666e-05,
+      "loss": 0.438,
+      "step": 1027
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.402459736888574e-05,
+      "loss": 0.4261,
+      "step": 1028
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4013425707191486e-05,
+      "loss": 0.4206,
+      "step": 1029
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.400224807156278e-05,
+      "loss": 0.458,
+      "step": 1030
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3991064478637397e-05,
+      "loss": 0.4696,
+      "step": 1031
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.397987494506198e-05,
+      "loss": 0.4771,
+      "step": 1032
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3968679487492001e-05,
+      "loss": 0.4447,
+      "step": 1033
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3957478122591773e-05,
+      "loss": 0.4578,
+      "step": 1034
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3946270867034377e-05,
+      "loss": 0.4743,
+      "step": 1035
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.393505773750168e-05,
+      "loss": 0.4831,
+      "step": 1036
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3923838750684287e-05,
+      "loss": 0.4655,
+      "step": 1037
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3912613923281517e-05,
+      "loss": 0.5334,
+      "step": 1038
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3901383272001389e-05,
+      "loss": 0.4655,
+      "step": 1039
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3890146813560592e-05,
+      "loss": 0.4327,
+      "step": 1040
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.387890456468445e-05,
+      "loss": 0.4438,
+      "step": 1041
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3867656542106911e-05,
+      "loss": 0.4794,
+      "step": 1042
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.385640276257052e-05,
+      "loss": 0.4575,
+      "step": 1043
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3845143242826384e-05,
+      "loss": 0.4546,
+      "step": 1044
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.383387799963417e-05,
+      "loss": 0.4931,
+      "step": 1045
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3822607049762038e-05,
+      "loss": 0.5538,
+      "step": 1046
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3811330409986667e-05,
+      "loss": 0.4617,
+      "step": 1047
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3800048097093193e-05,
+      "loss": 0.4975,
+      "step": 1048
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3788760127875198e-05,
+      "loss": 0.5099,
+      "step": 1049
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3777466519134686e-05,
+      "loss": 0.4487,
+      "step": 1050
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3766167287682052e-05,
+      "loss": 0.4308,
+      "step": 1051
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3754862450336065e-05,
+      "loss": 0.4541,
+      "step": 1052
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.374355202392383e-05,
+      "loss": 0.5018,
+      "step": 1053
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3732236025280783e-05,
+      "loss": 0.4374,
+      "step": 1054
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3720914471250644e-05,
+      "loss": 0.4624,
+      "step": 1055
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3709587378685404e-05,
+      "loss": 0.4663,
+      "step": 1056
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3698254764445307e-05,
+      "loss": 0.4189,
+      "step": 1057
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3686916645398802e-05,
+      "loss": 0.4802,
+      "step": 1058
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.367557303842254e-05,
+      "loss": 0.5007,
+      "step": 1059
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3664223960401345e-05,
+      "loss": 0.4742,
+      "step": 1060
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.365286942822817e-05,
+      "loss": 0.4654,
+      "step": 1061
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3641509458804104e-05,
+      "loss": 0.4877,
+      "step": 1062
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3630144069038319e-05,
+      "loss": 0.4471,
+      "step": 1063
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.361877327584805e-05,
+      "loss": 0.4744,
+      "step": 1064
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3607397096158588e-05,
+      "loss": 0.46,
+      "step": 1065
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3596015546903235e-05,
+      "loss": 0.4571,
+      "step": 1066
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3584628645023286e-05,
+      "loss": 0.4984,
+      "step": 1067
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3573236407468002e-05,
+      "loss": 0.4478,
+      "step": 1068
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3561838851194586e-05,
+      "loss": 0.4581,
+      "step": 1069
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3550435993168164e-05,
+      "loss": 0.4814,
+      "step": 1070
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3539027850361749e-05,
+      "loss": 0.4125,
+      "step": 1071
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3527614439756213e-05,
+      "loss": 0.4925,
+      "step": 1072
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3516195778340287e-05,
+      "loss": 0.4793,
+      "step": 1073
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3504771883110497e-05,
+      "loss": 0.4646,
+      "step": 1074
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3493342771071171e-05,
+      "loss": 0.4508,
+      "step": 1075
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.348190845923441e-05,
+      "loss": 0.4915,
+      "step": 1076
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3470468964620035e-05,
+      "loss": 0.4568,
+      "step": 1077
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3459024304255597e-05,
+      "loss": 0.4564,
+      "step": 1078
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.344757449517633e-05,
+      "loss": 0.4497,
+      "step": 1079
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3436119554425133e-05,
+      "loss": 0.4736,
+      "step": 1080
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.342465949905254e-05,
+      "loss": 0.4687,
+      "step": 1081
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3413194346116711e-05,
+      "loss": 0.4902,
+      "step": 1082
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3401724112683376e-05,
+      "loss": 0.5053,
+      "step": 1083
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3390248815825837e-05,
+      "loss": 0.4889,
+      "step": 1084
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.337876847262493e-05,
+      "loss": 0.4992,
+      "step": 1085
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3367283100169008e-05,
+      "loss": 0.4544,
+      "step": 1086
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3355792715553901e-05,
+      "loss": 0.4439,
+      "step": 1087
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.334429733588291e-05,
+      "loss": 0.4427,
+      "step": 1088
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3332796978266758e-05,
+      "loss": 0.4726,
+      "step": 1089
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3321291659823588e-05,
+      "loss": 0.4474,
+      "step": 1090
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3309781397678925e-05,
+      "loss": 0.5058,
+      "step": 1091
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3298266208965652e-05,
+      "loss": 0.4864,
+      "step": 1092
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.328674611082398e-05,
+      "loss": 0.4784,
+      "step": 1093
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3275221120401441e-05,
+      "loss": 0.4417,
+      "step": 1094
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3263691254852836e-05,
+      "loss": 0.4665,
+      "step": 1095
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3252156531340222e-05,
+      "loss": 0.4906,
+      "step": 1096
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3240616967032898e-05,
+      "loss": 0.4863,
+      "step": 1097
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.322907257910736e-05,
+      "loss": 0.454,
+      "step": 1098
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.321752338474729e-05,
+      "loss": 0.4241,
+      "step": 1099
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3205969401143517e-05,
+      "loss": 0.4414,
+      "step": 1100
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3194410645494005e-05,
+      "loss": 0.4798,
+      "step": 1101
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3182847135003815e-05,
+      "loss": 0.461,
+      "step": 1102
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3171278886885092e-05,
+      "loss": 0.4986,
+      "step": 1103
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.315970591835703e-05,
+      "loss": 0.4586,
+      "step": 1104
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.314812824664585e-05,
+      "loss": 0.4672,
+      "step": 1105
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3136545888984771e-05,
+      "loss": 0.4501,
+      "step": 1106
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3124958862613989e-05,
+      "loss": 0.4171,
+      "step": 1107
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.311336718478065e-05,
+      "loss": 0.4645,
+      "step": 1108
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3101770872738824e-05,
+      "loss": 0.4297,
+      "step": 1109
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 0.4576,
+      "step": 1110
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3078564415080446e-05,
+      "loss": 0.5218,
+      "step": 1111
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3066954304006418e-05,
+      "loss": 0.4345,
+      "step": 1112
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3055339627808898e-05,
+      "loss": 0.4473,
+      "step": 1113
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3043720403776194e-05,
+      "loss": 0.4557,
+      "step": 1114
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.303209664920337e-05,
+      "loss": 0.4747,
+      "step": 1115
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3020468381392244e-05,
+      "loss": 0.4663,
+      "step": 1116
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3008835617651347e-05,
+      "loss": 0.5223,
+      "step": 1117
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2997198375295905e-05,
+      "loss": 0.4847,
+      "step": 1118
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2985556671647807e-05,
+      "loss": 0.446,
+      "step": 1119
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2973910524035588e-05,
+      "loss": 0.5118,
+      "step": 1120
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2962259949794394e-05,
+      "loss": 0.4328,
+      "step": 1121
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2950604966265955e-05,
+      "loss": 0.4245,
+      "step": 1122
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.293894559079858e-05,
+      "loss": 0.5174,
+      "step": 1123
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2927281840747092e-05,
+      "loss": 0.4817,
+      "step": 1124
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2915613733472849e-05,
+      "loss": 0.4596,
+      "step": 1125
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.290394128634368e-05,
+      "loss": 0.4777,
+      "step": 1126
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.289226451673388e-05,
+      "loss": 0.4576,
+      "step": 1127
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.288058344202417e-05,
+      "loss": 0.4751,
+      "step": 1128
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.286889807960169e-05,
+      "loss": 0.4895,
+      "step": 1129
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.285720844685996e-05,
+      "loss": 0.4457,
+      "step": 1130
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.284551456119885e-05,
+      "loss": 0.4528,
+      "step": 1131
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.283381644002456e-05,
+      "loss": 0.4739,
+      "step": 1132
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2822114100749606e-05,
+      "loss": 0.4445,
+      "step": 1133
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2810407560792768e-05,
+      "loss": 0.4266,
+      "step": 1134
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.279869683757909e-05,
+      "loss": 0.4272,
+      "step": 1135
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.278698194853984e-05,
+      "loss": 0.4447,
+      "step": 1136
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2775262911112481e-05,
+      "loss": 0.4912,
+      "step": 1137
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2763539742740656e-05,
+      "loss": 0.4411,
+      "step": 1138
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.275181246087416e-05,
+      "loss": 0.4299,
+      "step": 1139
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.27400810829689e-05,
+      "loss": 0.5091,
+      "step": 1140
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.272834562648689e-05,
+      "loss": 0.427,
+      "step": 1141
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2716606108896215e-05,
+      "loss": 0.4567,
+      "step": 1142
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2704862547670999e-05,
+      "loss": 0.4863,
+      "step": 1143
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2693114960291383e-05,
+      "loss": 0.4462,
+      "step": 1144
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2681363364243511e-05,
+      "loss": 0.487,
+      "step": 1145
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2669607777019486e-05,
+      "loss": 0.4258,
+      "step": 1146
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2657848216117357e-05,
+      "loss": 0.4232,
+      "step": 1147
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2646084699041077e-05,
+      "loss": 0.4635,
+      "step": 1148
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2634317243300504e-05,
+      "loss": 0.5396,
+      "step": 1149
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2622545866411345e-05,
+      "loss": 0.4571,
+      "step": 1150
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2610770585895147e-05,
+      "loss": 0.4445,
+      "step": 1151
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2598991419279276e-05,
+      "loss": 0.4187,
+      "step": 1152
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2587208384096874e-05,
+      "loss": 0.4645,
+      "step": 1153
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2575421497886838e-05,
+      "loss": 0.4145,
+      "step": 1154
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2563630778193805e-05,
+      "loss": 0.4628,
+      "step": 1155
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2551836242568114e-05,
+      "loss": 0.518,
+      "step": 1156
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2540037908565786e-05,
+      "loss": 0.4799,
+      "step": 1157
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2528235793748497e-05,
+      "loss": 0.4733,
+      "step": 1158
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.251642991568355e-05,
+      "loss": 0.4502,
+      "step": 1159
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2504620291943841e-05,
+      "loss": 0.424,
+      "step": 1160
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2492806940107856e-05,
+      "loss": 0.4464,
+      "step": 1161
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2480989877759617e-05,
+      "loss": 0.456,
+      "step": 1162
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.246916912248868e-05,
+      "loss": 0.4596,
+      "step": 1163
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.245734469189009e-05,
+      "loss": 0.4766,
+      "step": 1164
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2445516603564365e-05,
+      "loss": 0.4668,
+      "step": 1165
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2433684875117469e-05,
+      "loss": 0.4762,
+      "step": 1166
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.242184952416078e-05,
+      "loss": 0.4971,
+      "step": 1167
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2410010568311081e-05,
+      "loss": 0.4731,
+      "step": 1168
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2398168025190502e-05,
+      "loss": 0.4425,
+      "step": 1169
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2386321912426524e-05,
+      "loss": 0.4589,
+      "step": 1170
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.237447224765194e-05,
+      "loss": 0.4619,
+      "step": 1171
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.236261904850483e-05,
+      "loss": 0.4214,
+      "step": 1172
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2350762332628527e-05,
+      "loss": 0.4507,
+      "step": 1173
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2338902117671618e-05,
+      "loss": 0.4854,
+      "step": 1174
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2327038421287879e-05,
+      "loss": 0.4553,
+      "step": 1175
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2315171261136272e-05,
+      "loss": 0.4972,
+      "step": 1176
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2303300654880926e-05,
+      "loss": 0.468,
+      "step": 1177
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2291426620191083e-05,
+      "loss": 0.5286,
+      "step": 1178
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.22795491747411e-05,
+      "loss": 0.4765,
+      "step": 1179
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2267668336210411e-05,
+      "loss": 0.4564,
+      "step": 1180
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2255784122283494e-05,
+      "loss": 0.4522,
+      "step": 1181
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2243896550649851e-05,
+      "loss": 0.4713,
+      "step": 1182
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2232005639003993e-05,
+      "loss": 0.4743,
+      "step": 1183
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2220111405045386e-05,
+      "loss": 0.4777,
+      "step": 1184
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2208213866478454e-05,
+      "loss": 0.4862,
+      "step": 1185
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2196313041012535e-05,
+      "loss": 0.4359,
+      "step": 1186
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2184408946361863e-05,
+      "loss": 0.4661,
+      "step": 1187
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2172501600245533e-05,
+      "loss": 0.479,
+      "step": 1188
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.216059102038748e-05,
+      "loss": 0.4722,
+      "step": 1189
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.214867722451646e-05,
+      "loss": 0.4687,
+      "step": 1190
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2136760230366007e-05,
+      "loss": 0.4814,
+      "step": 1191
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2124840055674419e-05,
+      "loss": 0.4779,
+      "step": 1192
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.211291671818473e-05,
+      "loss": 0.5335,
+      "step": 1193
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2100990235644679e-05,
+      "loss": 0.457,
+      "step": 1194
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2089060625806686e-05,
+      "loss": 0.4546,
+      "step": 1195
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2077127906427835e-05,
+      "loss": 0.5279,
+      "step": 1196
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.206519209526982e-05,
+      "loss": 0.4525,
+      "step": 1197
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2053253210098954e-05,
+      "loss": 0.4078,
+      "step": 1198
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2041311268686123e-05,
+      "loss": 0.4433,
+      "step": 1199
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.202936628880675e-05,
+      "loss": 0.4257,
+      "step": 1200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2017418288240796e-05,
+      "loss": 0.4985,
+      "step": 1201
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2005467284772714e-05,
+      "loss": 0.41,
+      "step": 1202
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1993513296191415e-05,
+      "loss": 0.465,
+      "step": 1203
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.198155634029027e-05,
+      "loss": 0.4741,
+      "step": 1204
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1969596434867063e-05,
+      "loss": 0.4148,
+      "step": 1205
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1957633597723957e-05,
+      "loss": 0.433,
+      "step": 1206
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1945667846667493e-05,
+      "loss": 0.4752,
+      "step": 1207
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1933699199508537e-05,
+      "loss": 0.4716,
+      "step": 1208
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.192172767406228e-05,
+      "loss": 0.4416,
+      "step": 1209
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1909753288148183e-05,
+      "loss": 0.4712,
+      "step": 1210
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1897776059589971e-05,
+      "loss": 0.4728,
+      "step": 1211
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.188579600621561e-05,
+      "loss": 0.5107,
+      "step": 1212
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.187381314585725e-05,
+      "loss": 0.5378,
+      "step": 1213
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.186182749635123e-05,
+      "loss": 0.5086,
+      "step": 1214
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.184983907553805e-05,
+      "loss": 0.47,
+      "step": 1215
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1837847901262316e-05,
+      "loss": 0.437,
+      "step": 1216
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1825853991372751e-05,
+      "loss": 0.5016,
+      "step": 1217
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1813857363722137e-05,
+      "loss": 0.4566,
+      "step": 1218
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1801858036167304e-05,
+      "loss": 0.4877,
+      "step": 1219
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1789856026569103e-05,
+      "loss": 0.4829,
+      "step": 1220
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1777851352792382e-05,
+      "loss": 0.4648,
+      "step": 1221
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1765844032705942e-05,
+      "loss": 0.4973,
+      "step": 1222
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1753834084182536e-05,
+      "loss": 0.498,
+      "step": 1223
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.174182152509882e-05,
+      "loss": 0.4847,
+      "step": 1224
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1729806373335337e-05,
+      "loss": 0.4084,
+      "step": 1225
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1717788646776498e-05,
+      "loss": 0.4832,
+      "step": 1226
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1705768363310535e-05,
+      "loss": 0.4416,
+      "step": 1227
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.16937455408295e-05,
+      "loss": 0.4442,
+      "step": 1228
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1681720197229201e-05,
+      "loss": 0.4416,
+      "step": 1229
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1669692350409223e-05,
+      "loss": 0.4846,
+      "step": 1230
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1657662018272863e-05,
+      "loss": 0.5888,
+      "step": 1231
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1645629218727124e-05,
+      "loss": 0.4819,
+      "step": 1232
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.163359396968268e-05,
+      "loss": 0.4324,
+      "step": 1233
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1621556289053846e-05,
+      "loss": 0.4815,
+      "step": 1234
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1609516194758562e-05,
+      "loss": 0.4322,
+      "step": 1235
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1597473704718359e-05,
+      "loss": 0.4774,
+      "step": 1236
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1585428836858338e-05,
+      "loss": 0.4153,
+      "step": 1237
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1573381609107128e-05,
+      "loss": 0.4753,
+      "step": 1238
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1561332039396886e-05,
+      "loss": 0.438,
+      "step": 1239
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1549280145663245e-05,
+      "loss": 0.4584,
+      "step": 1240
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1537225945845296e-05,
+      "loss": 0.4821,
+      "step": 1241
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1525169457885569e-05,
+      "loss": 0.4992,
+      "step": 1242
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1513110699729997e-05,
+      "loss": 0.4464,
+      "step": 1243
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1501049689327892e-05,
+      "loss": 0.4812,
+      "step": 1244
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.148898644463192e-05,
+      "loss": 0.4707,
+      "step": 1245
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1476920983598065e-05,
+      "loss": 0.4972,
+      "step": 1246
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1464853324185627e-05,
+      "loss": 0.4991,
+      "step": 1247
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1452783484357158e-05,
+      "loss": 0.4167,
+      "step": 1248
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.144071148207847e-05,
+      "loss": 0.4495,
+      "step": 1249
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.142863733531859e-05,
+      "loss": 0.4149,
+      "step": 1250
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1416561062049727e-05,
+      "loss": 0.4785,
+      "step": 1251
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1404482680247272e-05,
+      "loss": 0.5107,
+      "step": 1252
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.139240220788975e-05,
+      "loss": 0.478,
+      "step": 1253
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1380319662958785e-05,
+      "loss": 0.4369,
+      "step": 1254
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1368235063439103e-05,
+      "loss": 0.5198,
+      "step": 1255
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1356148427318478e-05,
+      "loss": 0.4992,
+      "step": 1256
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1344059772587716e-05,
+      "loss": 0.4334,
+      "step": 1257
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1331969117240632e-05,
+      "loss": 0.5036,
+      "step": 1258
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1319876479274017e-05,
+      "loss": 0.4529,
+      "step": 1259
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1307781876687611e-05,
+      "loss": 0.4722,
+      "step": 1260
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1295685327484084e-05,
+      "loss": 0.4387,
+      "step": 1261
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1283586849668997e-05,
+      "loss": 0.4589,
+      "step": 1262
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1271486461250782e-05,
+      "loss": 0.4447,
+      "step": 1263
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.125938418024072e-05,
+      "loss": 0.4333,
+      "step": 1264
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.124728002465291e-05,
+      "loss": 0.4795,
+      "step": 1265
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.123517401250423e-05,
+      "loss": 0.5002,
+      "step": 1266
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1223066161814333e-05,
+      "loss": 0.5032,
+      "step": 1267
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1210956490605604e-05,
+      "loss": 0.3928,
+      "step": 1268
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1198845016903144e-05,
+      "loss": 0.4894,
+      "step": 1269
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1186731758734722e-05,
+      "loss": 0.4756,
+      "step": 1270
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1174616734130779e-05,
+      "loss": 0.4691,
+      "step": 1271
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.116249996112438e-05,
+      "loss": 0.4161,
+      "step": 1272
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1150381457751184e-05,
+      "loss": 0.4858,
+      "step": 1273
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1138261242049443e-05,
+      "loss": 0.4601,
+      "step": 1274
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.112613933205994e-05,
+      "loss": 0.4345,
+      "step": 1275
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1114015745825993e-05,
+      "loss": 0.468,
+      "step": 1276
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1101890501393407e-05,
+      "loss": 0.4647,
+      "step": 1277
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.108976361681046e-05,
+      "loss": 0.4577,
+      "step": 1278
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1077635110127867e-05,
+      "loss": 0.4485,
+      "step": 1279
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1065504999398762e-05,
+      "loss": 0.477,
+      "step": 1280
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1053373302678665e-05,
+      "loss": 0.4574,
+      "step": 1281
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1041240038025458e-05,
+      "loss": 0.4088,
+      "step": 1282
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1029105223499348e-05,
+      "loss": 0.4287,
+      "step": 1283
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1016968877162863e-05,
+      "loss": 0.4541,
+      "step": 1284
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1004831017080802e-05,
+      "loss": 0.4801,
+      "step": 1285
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0992691661320216e-05,
+      "loss": 0.532,
+      "step": 1286
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0980550827950394e-05,
+      "loss": 0.4477,
+      "step": 1287
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.096840853504281e-05,
+      "loss": 0.445,
+      "step": 1288
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0956264800671112e-05,
+      "loss": 0.5075,
+      "step": 1289
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0944119642911108e-05,
+      "loss": 0.4453,
+      "step": 1290
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0931973079840712e-05,
+      "loss": 0.4228,
+      "step": 1291
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.091982512953993e-05,
+      "loss": 0.476,
+      "step": 1292
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0907675810090836e-05,
+      "loss": 0.4702,
+      "step": 1293
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0895525139577546e-05,
+      "loss": 0.4789,
+      "step": 1294
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0883373136086173e-05,
+      "loss": 0.4613,
+      "step": 1295
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0871219817704832e-05,
+      "loss": 0.4518,
+      "step": 1296
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0859065202523584e-05,
+      "loss": 0.4647,
+      "step": 1297
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0846909308634426e-05,
+      "loss": 0.5016,
+      "step": 1298
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0834752154131252e-05,
+      "loss": 0.4759,
+      "step": 1299
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0822593757109835e-05,
+      "loss": 0.4329,
+      "step": 1300
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0810434135667802e-05,
+      "loss": 0.4322,
+      "step": 1301
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0798273307904595e-05,
+      "loss": 0.5154,
+      "step": 1302
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0786111291921462e-05,
+      "loss": 0.4221,
+      "step": 1303
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0773948105821405e-05,
+      "loss": 0.463,
+      "step": 1304
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0761783767709182e-05,
+      "loss": 0.4851,
+      "step": 1305
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0749618295691255e-05,
+      "loss": 0.453,
+      "step": 1306
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.073745170787578e-05,
+      "loss": 0.4718,
+      "step": 1307
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0725284022372575e-05,
+      "loss": 0.4717,
+      "step": 1308
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0713115257293087e-05,
+      "loss": 0.5417,
+      "step": 1309
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0700945430750373e-05,
+      "loss": 0.498,
+      "step": 1310
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0688774560859065e-05,
+      "loss": 0.4441,
+      "step": 1311
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0676602665735356e-05,
+      "loss": 0.4718,
+      "step": 1312
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0664429763496964e-05,
+      "loss": 0.4774,
+      "step": 1313
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0652255872263097e-05,
+      "loss": 0.5082,
+      "step": 1314
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0640081010154444e-05,
+      "loss": 0.4682,
+      "step": 1315
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0627905195293135e-05,
+      "loss": 0.4787,
+      "step": 1316
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0615728445802723e-05,
+      "loss": 0.4483,
+      "step": 1317
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0603550779808143e-05,
+      "loss": 0.4609,
+      "step": 1318
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0591372215435707e-05,
+      "loss": 0.432,
+      "step": 1319
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0579192770813053e-05,
+      "loss": 0.417,
+      "step": 1320
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0567012464069135e-05,
+      "loss": 0.4553,
+      "step": 1321
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0554831313334188e-05,
+      "loss": 0.4314,
+      "step": 1322
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0542649336739704e-05,
+      "loss": 0.4905,
+      "step": 1323
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0530466552418402e-05,
+      "loss": 0.4627,
+      "step": 1324
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0518282978504209e-05,
+      "loss": 0.4231,
+      "step": 1325
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0506098633132224e-05,
+      "loss": 0.501,
+      "step": 1326
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.049391353443869e-05,
+      "loss": 0.4965,
+      "step": 1327
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.048172770056098e-05,
+      "loss": 0.4431,
+      "step": 1328
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0469541149637554e-05,
+      "loss": 0.4804,
+      "step": 1329
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0457353899807947e-05,
+      "loss": 0.4616,
+      "step": 1330
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.044516596921272e-05,
+      "loss": 0.4523,
+      "step": 1331
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0432977375993467e-05,
+      "loss": 0.4497,
+      "step": 1332
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0420788138292751e-05,
+      "loss": 0.4559,
+      "step": 1333
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0408598274254103e-05,
+      "loss": 0.5028,
+      "step": 1334
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0396407802021986e-05,
+      "loss": 0.4568,
+      "step": 1335
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0384216739741766e-05,
+      "loss": 0.453,
+      "step": 1336
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0372025105559689e-05,
+      "loss": 0.4727,
+      "step": 1337
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.035983291762285e-05,
+      "loss": 0.4889,
+      "step": 1338
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0347640194079169e-05,
+      "loss": 0.4512,
+      "step": 1339
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0335446953077366e-05,
+      "loss": 0.4541,
+      "step": 1340
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.032325321276693e-05,
+      "loss": 0.4433,
+      "step": 1341
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0311058991298093e-05,
+      "loss": 0.4699,
+      "step": 1342
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0298864306821797e-05,
+      "loss": 0.4858,
+      "step": 1343
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0286669177489685e-05,
+      "loss": 0.44,
+      "step": 1344
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.027447362145405e-05,
+      "loss": 0.5006,
+      "step": 1345
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0262277656867834e-05,
+      "loss": 0.4367,
+      "step": 1346
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0250081301884574e-05,
+      "loss": 0.4265,
+      "step": 1347
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.023788457465839e-05,
+      "loss": 0.4365,
+      "step": 1348
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0225687493343966e-05,
+      "loss": 0.454,
+      "step": 1349
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0213490076096502e-05,
+      "loss": 0.4083,
+      "step": 1350
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0201292341071704e-05,
+      "loss": 0.5059,
+      "step": 1351
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0189094306425755e-05,
+      "loss": 0.4705,
+      "step": 1352
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0176895990315267e-05,
+      "loss": 0.4642,
+      "step": 1353
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0164697410897293e-05,
+      "loss": 0.4616,
+      "step": 1354
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.015249858632926e-05,
+      "loss": 0.4591,
+      "step": 1355
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0140299534768972e-05,
+      "loss": 0.4275,
+      "step": 1356
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0128100274374563e-05,
+      "loss": 0.5061,
+      "step": 1357
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0115900823304486e-05,
+      "loss": 0.4482,
+      "step": 1358
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0103701199717466e-05,
+      "loss": 0.4584,
+      "step": 1359
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0091501421772496e-05,
+      "loss": 0.466,
+      "step": 1360
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0079301507628791e-05,
+      "loss": 0.4583,
+      "step": 1361
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.006710147544578e-05,
+      "loss": 0.4661,
+      "step": 1362
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.005490134338305e-05,
+      "loss": 0.4525,
+      "step": 1363
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.004270112960035e-05,
+      "loss": 0.4966,
+      "step": 1364
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0030500852257545e-05,
+      "loss": 0.4318,
+      "step": 1365
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0018300529514604e-05,
+      "loss": 0.4633,
+      "step": 1366
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0006100179531548e-05,
+      "loss": 0.4456,
+      "step": 1367
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.993899820468454e-06,
+      "loss": 0.4916,
+      "step": 1368
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.9816994704854e-06,
+      "loss": 0.454,
+      "step": 1369
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.969499147742455e-06,
+      "loss": 0.4436,
+      "step": 1370
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.957298870399657e-06,
+      "loss": 0.4552,
+      "step": 1371
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.945098656616954e-06,
+      "loss": 0.419,
+      "step": 1372
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.932898524554225e-06,
+      "loss": 0.4967,
+      "step": 1373
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.92069849237121e-06,
+      "loss": 0.4148,
+      "step": 1374
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.908498578227505e-06,
+      "loss": 0.4508,
+      "step": 1375
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.896298800282534e-06,
+      "loss": 0.4492,
+      "step": 1376
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.884099176695518e-06,
+      "loss": 0.4742,
+      "step": 1377
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.871899725625438e-06,
+      "loss": 0.4713,
+      "step": 1378
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.859700465231032e-06,
+      "loss": 0.4688,
+      "step": 1379
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.847501413670742e-06,
+      "loss": 0.4592,
+      "step": 1380
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.83530258910271e-06,
+      "loss": 0.4527,
+      "step": 1381
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.823104009684733e-06,
+      "loss": 0.4563,
+      "step": 1382
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.810905693574248e-06,
+      "loss": 0.442,
+      "step": 1383
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.798707658928298e-06,
+      "loss": 0.5632,
+      "step": 1384
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.786509923903503e-06,
+      "loss": 0.4144,
+      "step": 1385
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.774312506656039e-06,
+      "loss": 0.427,
+      "step": 1386
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.762115425341613e-06,
+      "loss": 0.4796,
+      "step": 1387
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.74991869811543e-06,
+      "loss": 0.4443,
+      "step": 1388
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.73772234313217e-06,
+      "loss": 0.4619,
+      "step": 1389
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.725526378545953e-06,
+      "loss": 0.479,
+      "step": 1390
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.71333082251032e-06,
+      "loss": 0.4185,
+      "step": 1391
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.701135693178205e-06,
+      "loss": 0.4572,
+      "step": 1392
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.68894100870191e-06,
+      "loss": 0.4465,
+      "step": 1393
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.676746787233071e-06,
+      "loss": 0.4378,
+      "step": 1394
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.664553046922634e-06,
+      "loss": 0.4422,
+      "step": 1395
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.652359805920835e-06,
+      "loss": 0.4063,
+      "step": 1396
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.640167082377155e-06,
+      "loss": 0.4572,
+      "step": 1397
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.627974894440315e-06,
+      "loss": 0.4625,
+      "step": 1398
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.615783260258236e-06,
+      "loss": 0.4654,
+      "step": 1399
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.603592197978017e-06,
+      "loss": 0.4783,
+      "step": 1400
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.591401725745898e-06,
+      "loss": 0.4329,
+      "step": 1401
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.57921186170725e-06,
+      "loss": 0.4589,
+      "step": 1402
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.567022624006538e-06,
+      "loss": 0.4627,
+      "step": 1403
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.554834030787283e-06,
+      "loss": 0.4437,
+      "step": 1404
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.542646100192056e-06,
+      "loss": 0.4347,
+      "step": 1405
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.530458850362447e-06,
+      "loss": 0.4522,
+      "step": 1406
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.51827229943902e-06,
+      "loss": 0.4779,
+      "step": 1407
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.50608646556131e-06,
+      "loss": 0.4188,
+      "step": 1408
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.493901366867781e-06,
+      "loss": 0.4445,
+      "step": 1409
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.481717021495795e-06,
+      "loss": 0.4372,
+      "step": 1410
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.469533447581603e-06,
+      "loss": 0.4705,
+      "step": 1411
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.457350663260301e-06,
+      "loss": 0.4789,
+      "step": 1412
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.445168686665814e-06,
+      "loss": 0.4911,
+      "step": 1413
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.432987535930865e-06,
+      "loss": 0.4274,
+      "step": 1414
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.42080722918695e-06,
+      "loss": 0.4061,
+      "step": 1415
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.408627784564295e-06,
+      "loss": 0.4705,
+      "step": 1416
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.39644922019186e-06,
+      "loss": 0.4421,
+      "step": 1417
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.38427155419728e-06,
+      "loss": 0.4372,
+      "step": 1418
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.372094804706867e-06,
+      "loss": 0.404,
+      "step": 1419
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.359918989845558e-06,
+      "loss": 0.4485,
+      "step": 1420
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.347744127736906e-06,
+      "loss": 0.4513,
+      "step": 1421
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.33557023650304e-06,
+      "loss": 0.4604,
+      "step": 1422
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.323397334264646e-06,
+      "loss": 0.4053,
+      "step": 1423
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.311225439140938e-06,
+      "loss": 0.4023,
+      "step": 1424
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.299054569249629e-06,
+      "loss": 0.5111,
+      "step": 1425
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.286884742706913e-06,
+      "loss": 0.4325,
+      "step": 1426
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.274715977627425e-06,
+      "loss": 0.4505,
+      "step": 1427
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.262548292124224e-06,
+      "loss": 0.4863,
+      "step": 1428
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.250381704308748e-06,
+      "loss": 0.4606,
+      "step": 1429
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.238216232290821e-06,
+      "loss": 0.4742,
+      "step": 1430
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.226051894178597e-06,
+      "loss": 0.4756,
+      "step": 1431
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.213888708078543e-06,
+      "loss": 0.4395,
+      "step": 1432
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.201726692095405e-06,
+      "loss": 0.4178,
+      "step": 1433
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.189565864332205e-06,
+      "loss": 0.4146,
+      "step": 1434
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.177406242890168e-06,
+      "loss": 0.4185,
+      "step": 1435
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.165247845868751e-06,
+      "loss": 0.4786,
+      "step": 1436
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.153090691365576e-06,
+      "loss": 0.4589,
+      "step": 1437
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.140934797476418e-06,
+      "loss": 0.4559,
+      "step": 1438
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.12878018229517e-06,
+      "loss": 0.4423,
+      "step": 1439
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.116626863913827e-06,
+      "loss": 0.4305,
+      "step": 1440
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.10447486042246e-06,
+      "loss": 0.481,
+      "step": 1441
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.092324189909166e-06,
+      "loss": 0.4393,
+      "step": 1442
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.080174870460075e-06,
+      "loss": 0.4964,
+      "step": 1443
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.06802692015929e-06,
+      "loss": 0.4952,
+      "step": 1444
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.055880357088892e-06,
+      "loss": 0.4594,
+      "step": 1445
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.043735199328887e-06,
+      "loss": 0.447,
+      "step": 1446
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.031591464957197e-06,
+      "loss": 0.4541,
+      "step": 1447
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.01944917204961e-06,
+      "loss": 0.4944,
+      "step": 1448
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.007308338679786e-06,
+      "loss": 0.4834,
+      "step": 1449
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.995168982919203e-06,
+      "loss": 0.4497,
+      "step": 1450
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.98303112283714e-06,
+      "loss": 0.4661,
+      "step": 1451
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.970894776500652e-06,
+      "loss": 0.4599,
+      "step": 1452
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.958759961974548e-06,
+      "loss": 0.4482,
+      "step": 1453
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.946626697321337e-06,
+      "loss": 0.4669,
+      "step": 1454
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.934495000601241e-06,
+      "loss": 0.478,
+      "step": 1455
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.922364889872134e-06,
+      "loss": 0.4826,
+      "step": 1456
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.910236383189542e-06,
+      "loss": 0.4428,
+      "step": 1457
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.898109498606595e-06,
+      "loss": 0.4913,
+      "step": 1458
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.885984254174008e-06,
+      "loss": 0.4825,
+      "step": 1459
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.873860667940066e-06,
+      "loss": 0.4805,
+      "step": 1460
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.861738757950562e-06,
+      "loss": 0.476,
+      "step": 1461
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.849618542248817e-06,
+      "loss": 0.4121,
+      "step": 1462
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.837500038875624e-06,
+      "loss": 0.4732,
+      "step": 1463
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.825383265869223e-06,
+      "loss": 0.4716,
+      "step": 1464
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.81326824126528e-06,
+      "loss": 0.4681,
+      "step": 1465
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.801154983096863e-06,
+      "loss": 0.4311,
+      "step": 1466
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.789043509394397e-06,
+      "loss": 0.4903,
+      "step": 1467
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.776933838185669e-06,
+      "loss": 0.4504,
+      "step": 1468
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.764825987495772e-06,
+      "loss": 0.4565,
+      "step": 1469
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.752719975347094e-06,
+      "loss": 0.4182,
+      "step": 1470
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.74061581975928e-06,
+      "loss": 0.4798,
+      "step": 1471
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.728513538749223e-06,
+      "loss": 0.5097,
+      "step": 1472
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.716413150331008e-06,
+      "loss": 0.4608,
+      "step": 1473
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.704314672515919e-06,
+      "loss": 0.4762,
+      "step": 1474
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.692218123312392e-06,
+      "loss": 0.4702,
+      "step": 1475
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.680123520725986e-06,
+      "loss": 0.4421,
+      "step": 1476
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.66803088275937e-06,
+      "loss": 0.4934,
+      "step": 1477
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.655940227412289e-06,
+      "loss": 0.4683,
+      "step": 1478
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.643851572681528e-06,
+      "loss": 0.4171,
+      "step": 1479
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.6317649365609e-06,
+      "loss": 0.4736,
+      "step": 1480
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.619680337041217e-06,
+      "loss": 0.4731,
+      "step": 1481
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.607597792110254e-06,
+      "loss": 0.4596,
+      "step": 1482
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.595517319752728e-06,
+      "loss": 0.4412,
+      "step": 1483
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.583438937950273e-06,
+      "loss": 0.4451,
+      "step": 1484
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.571362664681416e-06,
+      "loss": 0.4506,
+      "step": 1485
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.559288517921533e-06,
+      "loss": 0.4419,
+      "step": 1486
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.547216515642845e-06,
+      "loss": 0.5051,
+      "step": 1487
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.535146675814376e-06,
+      "loss": 0.4665,
+      "step": 1488
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.523079016401933e-06,
+      "loss": 0.4258,
+      "step": 1489
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.511013555368081e-06,
+      "loss": 0.4816,
+      "step": 1490
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.498950310672111e-06,
+      "loss": 0.4411,
+      "step": 1491
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.486889300270007e-06,
+      "loss": 0.4827,
+      "step": 1492
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.474830542114435e-06,
+      "loss": 0.4843,
+      "step": 1493
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.462774054154708e-06,
+      "loss": 0.5263,
+      "step": 1494
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.450719854336759e-06,
+      "loss": 0.4069,
+      "step": 1495
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.438667960603117e-06,
+      "loss": 0.4534,
+      "step": 1496
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.426618390892875e-06,
+      "loss": 0.4323,
+      "step": 1497
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.41457116314167e-06,
+      "loss": 0.4289,
+      "step": 1498
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.402526295281644e-06,
+      "loss": 0.4276,
+      "step": 1499
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.390483805241442e-06,
+      "loss": 0.4958,
+      "step": 1500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.378443710946157e-06,
+      "loss": 0.4944,
+      "step": 1501
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.366406030317325e-06,
+      "loss": 0.4838,
+      "step": 1502
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.354370781272877e-06,
+      "loss": 0.4679,
+      "step": 1503
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.342337981727142e-06,
+      "loss": 0.46,
+      "step": 1504
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.330307649590782e-06,
+      "loss": 0.4623,
+      "step": 1505
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.3182798027708e-06,
+      "loss": 0.4658,
+      "step": 1506
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.306254459170504e-06,
+      "loss": 0.4844,
+      "step": 1507
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.294231636689465e-06,
+      "loss": 0.4591,
+      "step": 1508
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.282211353223503e-06,
+      "loss": 0.48,
+      "step": 1509
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.270193626664666e-06,
+      "loss": 0.4192,
+      "step": 1510
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.258178474901187e-06,
+      "loss": 0.4729,
+      "step": 1511
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.246165915817467e-06,
+      "loss": 0.4792,
+      "step": 1512
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.234155967294062e-06,
+      "loss": 0.4614,
+      "step": 1513
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.222148647207623e-06,
+      "loss": 0.4707,
+      "step": 1514
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.210143973430897e-06,
+      "loss": 0.4498,
+      "step": 1515
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.198141963832701e-06,
+      "loss": 0.4358,
+      "step": 1516
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.186142636277868e-06,
+      "loss": 0.489,
+      "step": 1517
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.174146008627252e-06,
+      "loss": 0.4753,
+      "step": 1518
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.162152098737685e-06,
+      "loss": 0.4288,
+      "step": 1519
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.150160924461954e-06,
+      "loss": 0.4666,
+      "step": 1520
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.13817250364877e-06,
+      "loss": 0.4578,
+      "step": 1521
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.126186854142752e-06,
+      "loss": 0.4333,
+      "step": 1522
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.114203993784395e-06,
+      "loss": 0.4198,
+      "step": 1523
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.10222394041003e-06,
+      "loss": 0.4065,
+      "step": 1524
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.09024671185182e-06,
+      "loss": 0.4165,
+      "step": 1525
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.078272325937723e-06,
+      "loss": 0.4453,
+      "step": 1526
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.066300800491463e-06,
+      "loss": 0.4338,
+      "step": 1527
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.05433215333251e-06,
+      "loss": 0.4976,
+      "step": 1528
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.042366402276047e-06,
+      "loss": 0.5136,
+      "step": 1529
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.030403565132942e-06,
+      "loss": 0.4529,
+      "step": 1530
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.018443659709731e-06,
+      "loss": 0.4443,
+      "step": 1531
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.006486703808587e-06,
+      "loss": 0.4041,
+      "step": 1532
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.99453271522729e-06,
+      "loss": 0.4289,
+      "step": 1533
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.982581711759206e-06,
+      "loss": 0.4356,
+      "step": 1534
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.970633711193253e-06,
+      "loss": 0.491,
+      "step": 1535
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.958688731313884e-06,
+      "loss": 0.4395,
+      "step": 1536
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.946746789901048e-06,
+      "loss": 0.4836,
+      "step": 1537
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.934807904730182e-06,
+      "loss": 0.4165,
+      "step": 1538
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.922872093572167e-06,
+      "loss": 0.4547,
+      "step": 1539
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.910939374193314e-06,
+      "loss": 0.441,
+      "step": 1540
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.899009764355321e-06,
+      "loss": 0.4645,
+      "step": 1541
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.887083281815274e-06,
+      "loss": 0.4633,
+      "step": 1542
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.875159944325582e-06,
+      "loss": 0.431,
+      "step": 1543
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.863239769633995e-06,
+      "loss": 0.4881,
+      "step": 1544
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.851322775483543e-06,
+      "loss": 0.4714,
+      "step": 1545
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.839408979612521e-06,
+      "loss": 0.4526,
+      "step": 1546
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.827498399754469e-06,
+      "loss": 0.4473,
+      "step": 1547
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.81559105363814e-06,
+      "loss": 0.4824,
+      "step": 1548
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.803686958987468e-06,
+      "loss": 0.4887,
+      "step": 1549
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.791786133521548e-06,
+      "loss": 0.4705,
+      "step": 1550
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.77988859495462e-06,
+      "loss": 0.442,
+      "step": 1551
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.767994360996012e-06,
+      "loss": 0.4052,
+      "step": 1552
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.75610344935015e-06,
+      "loss": 0.4326,
+      "step": 1553
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.744215877716511e-06,
+      "loss": 0.4189,
+      "step": 1554
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.732331663789592e-06,
+      "loss": 0.424,
+      "step": 1555
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.720450825258903e-06,
+      "loss": 0.4231,
+      "step": 1556
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.70857337980892e-06,
+      "loss": 0.4081,
+      "step": 1557
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.696699345119078e-06,
+      "loss": 0.431,
+      "step": 1558
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.684828738863727e-06,
+      "loss": 0.4524,
+      "step": 1559
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.672961578712126e-06,
+      "loss": 0.4851,
+      "step": 1560
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.661097882328384e-06,
+      "loss": 0.3979,
+      "step": 1561
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.649237667371475e-06,
+      "loss": 0.4344,
+      "step": 1562
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.637380951495175e-06,
+      "loss": 0.4922,
+      "step": 1563
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.625527752348062e-06,
+      "loss": 0.5474,
+      "step": 1564
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.613678087573476e-06,
+      "loss": 0.4615,
+      "step": 1565
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.601831974809499e-06,
+      "loss": 0.4608,
+      "step": 1566
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.589989431688922e-06,
+      "loss": 0.4511,
+      "step": 1567
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.578150475839221e-06,
+      "loss": 0.4269,
+      "step": 1568
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.566315124882535e-06,
+      "loss": 0.4415,
+      "step": 1569
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.554483396435638e-06,
+      "loss": 0.5114,
+      "step": 1570
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5426553081099126e-06,
+      "loss": 0.4758,
+      "step": 1571
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.530830877511323e-06,
+      "loss": 0.5116,
+      "step": 1572
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.519010122240389e-06,
+      "loss": 0.4925,
+      "step": 1573
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5071930598921495e-06,
+      "loss": 0.4775,
+      "step": 1574
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.495379708056162e-06,
+      "loss": 0.469,
+      "step": 1575
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.483570084316454e-06,
+      "loss": 0.4129,
+      "step": 1576
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.471764206251504e-06,
+      "loss": 0.4237,
+      "step": 1577
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.459962091434214e-06,
+      "loss": 0.4361,
+      "step": 1578
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.448163757431891e-06,
+      "loss": 0.4106,
+      "step": 1579
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.436369221806201e-06,
+      "loss": 0.4633,
+      "step": 1580
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.424578502113166e-06,
+      "loss": 0.4344,
+      "step": 1581
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4127916159031296e-06,
+      "loss": 0.497,
+      "step": 1582
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.401008580720725e-06,
+      "loss": 0.4424,
+      "step": 1583
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.389229414104854e-06,
+      "loss": 0.472,
+      "step": 1584
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.377454133588657e-06,
+      "loss": 0.4258,
+      "step": 1585
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.365682756699501e-06,
+      "loss": 0.458,
+      "step": 1586
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.353915300958927e-06,
+      "loss": 0.4506,
+      "step": 1587
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.342151783882647e-06,
+      "loss": 0.4014,
+      "step": 1588
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.330392222980516e-06,
+      "loss": 0.4502,
+      "step": 1589
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.318636635756491e-06,
+      "loss": 0.4664,
+      "step": 1590
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3068850397086175e-06,
+      "loss": 0.4524,
+      "step": 1591
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.295137452329006e-06,
+      "loss": 0.4139,
+      "step": 1592
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.283393891103787e-06,
+      "loss": 0.4914,
+      "step": 1593
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.271654373513112e-06,
+      "loss": 0.4362,
+      "step": 1594
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.259918917031103e-06,
+      "loss": 0.5008,
+      "step": 1595
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.248187539125844e-06,
+      "loss": 0.4379,
+      "step": 1596
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.236460257259343e-06,
+      "loss": 0.4195,
+      "step": 1597
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.224737088887523e-06,
+      "loss": 0.4674,
+      "step": 1598
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.213018051460163e-06,
+      "loss": 0.4463,
+      "step": 1599
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.201303162420914e-06,
+      "loss": 0.4439,
+      "step": 1600
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.189592439207236e-06,
+      "loss": 0.402,
+      "step": 1601
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1778858992503975e-06,
+      "loss": 0.4486,
+      "step": 1602
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.166183559975442e-06,
+      "loss": 0.4996,
+      "step": 1603
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.154485438801153e-06,
+      "loss": 0.4385,
+      "step": 1604
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.142791553140045e-06,
+      "loss": 0.4582,
+      "step": 1605
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.131101920398311e-06,
+      "loss": 0.4705,
+      "step": 1606
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1194165579758326e-06,
+      "loss": 0.4798,
+      "step": 1607
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.107735483266122e-06,
+      "loss": 0.4481,
+      "step": 1608
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.096058713656322e-06,
+      "loss": 0.4664,
+      "step": 1609
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.084386266527152e-06,
+      "loss": 0.4552,
+      "step": 1610
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.072718159252912e-06,
+      "loss": 0.4758,
+      "step": 1611
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.061054409201426e-06,
+      "loss": 0.4795,
+      "step": 1612
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.049395033734045e-06,
+      "loss": 0.4989,
+      "step": 1613
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.037740050205608e-06,
+      "loss": 0.4641,
+      "step": 1614
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.026089475964415e-06,
+      "loss": 0.4211,
+      "step": 1615
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.014443328352194e-06,
+      "loss": 0.4406,
+      "step": 1616
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0028016247041004e-06,
+      "loss": 0.429,
+      "step": 1617
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.991164382348657e-06,
+      "loss": 0.4208,
+      "step": 1618
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.97953161860776e-06,
+      "loss": 0.4055,
+      "step": 1619
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.967903350796632e-06,
+      "loss": 0.4207,
+      "step": 1620
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.95627959622381e-06,
+      "loss": 0.4221,
+      "step": 1621
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.944660372191102e-06,
+      "loss": 0.4133,
+      "step": 1622
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.933045695993583e-06,
+      "loss": 0.4436,
+      "step": 1623
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.921435584919559e-06,
+      "loss": 0.4236,
+      "step": 1624
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 0.4784,
+      "step": 1625
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.89822912726118e-06,
+      "loss": 0.4248,
+      "step": 1626
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8866328152193514e-06,
+      "loss": 0.4608,
+      "step": 1627
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.875041137386011e-06,
+      "loss": 0.4775,
+      "step": 1628
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8634541110152285e-06,
+      "loss": 0.4587,
+      "step": 1629
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.851871753354154e-06,
+      "loss": 0.4296,
+      "step": 1630
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.84029408164297e-06,
+      "loss": 0.4264,
+      "step": 1631
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.828721113114911e-06,
+      "loss": 0.4544,
+      "step": 1632
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8171528649961885e-06,
+      "loss": 0.4835,
+      "step": 1633
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.805589354505997e-06,
+      "loss": 0.4168,
+      "step": 1634
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.794030598856484e-06,
+      "loss": 0.4484,
+      "step": 1635
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.782476615252713e-06,
+      "loss": 0.4461,
+      "step": 1636
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.770927420892644e-06,
+      "loss": 0.4905,
+      "step": 1637
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.759383032967106e-06,
+      "loss": 0.4214,
+      "step": 1638
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.747843468659782e-06,
+      "loss": 0.4333,
+      "step": 1639
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.736308745147169e-06,
+      "loss": 0.4647,
+      "step": 1640
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7247788795985615e-06,
+      "loss": 0.4311,
+      "step": 1641
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.713253889176019e-06,
+      "loss": 0.4404,
+      "step": 1642
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.701733791034353e-06,
+      "loss": 0.4482,
+      "step": 1643
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.690218602321078e-06,
+      "loss": 0.4444,
+      "step": 1644
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.678708340176414e-06,
+      "loss": 0.4487,
+      "step": 1645
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.667203021733244e-06,
+      "loss": 0.4155,
+      "step": 1646
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.655702664117094e-06,
+      "loss": 0.4177,
+      "step": 1647
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.644207284446099e-06,
+      "loss": 0.4534,
+      "step": 1648
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6327168998309965e-06,
+      "loss": 0.4239,
+      "step": 1649
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.621231527375071e-06,
+      "loss": 0.4411,
+      "step": 1650
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.609751184174167e-06,
+      "loss": 0.4504,
+      "step": 1651
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.598275887316626e-06,
+      "loss": 0.4296,
+      "step": 1652
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.586805653883292e-06,
+      "loss": 0.487,
+      "step": 1653
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.575340500947459e-06,
+      "loss": 0.4658,
+      "step": 1654
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.563880445574873e-06,
+      "loss": 0.447,
+      "step": 1655
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.552425504823674e-06,
+      "loss": 0.4467,
+      "step": 1656
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5409756957444075e-06,
+      "loss": 0.4319,
+      "step": 1657
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.529531035379969e-06,
+      "loss": 0.4179,
+      "step": 1658
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.518091540765595e-06,
+      "loss": 0.4038,
+      "step": 1659
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.506657228928828e-06,
+      "loss": 0.434,
+      "step": 1660
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.495228116889509e-06,
+      "loss": 0.458,
+      "step": 1661
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.483804221659719e-06,
+      "loss": 0.4724,
+      "step": 1662
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.472385560243788e-06,
+      "loss": 0.3936,
+      "step": 1663
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.460972149638256e-06,
+      "loss": 0.4319,
+      "step": 1664
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4495640068318365e-06,
+      "loss": 0.4423,
+      "step": 1665
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.438161148805414e-06,
+      "loss": 0.4274,
+      "step": 1666
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4267635925319995e-06,
+      "loss": 0.4553,
+      "step": 1667
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4153713549767184e-06,
+      "loss": 0.3773,
+      "step": 1668
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.403984453096768e-06,
+      "loss": 0.436,
+      "step": 1669
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.392602903841416e-06,
+      "loss": 0.4414,
+      "step": 1670
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.381226724151953e-06,
+      "loss": 0.4702,
+      "step": 1671
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.369855930961686e-06,
+      "loss": 0.4479,
+      "step": 1672
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.358490541195899e-06,
+      "loss": 0.4562,
+      "step": 1673
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3471305717718315e-06,
+      "loss": 0.4174,
+      "step": 1674
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.33577603959866e-06,
+      "loss": 0.4295,
+      "step": 1675
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.324426961577461e-06,
+      "loss": 0.4423,
+      "step": 1676
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3130833546012e-06,
+      "loss": 0.4046,
+      "step": 1677
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.301745235554695e-06,
+      "loss": 0.4016,
+      "step": 1678
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.290412621314596e-06,
+      "loss": 0.4627,
+      "step": 1679
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2790855287493605e-06,
+      "loss": 0.4561,
+      "step": 1680
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.267763974719222e-06,
+      "loss": 0.4588,
+      "step": 1681
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.256447976076172e-06,
+      "loss": 0.4796,
+      "step": 1682
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.245137549663938e-06,
+      "loss": 0.4481,
+      "step": 1683
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.233832712317949e-06,
+      "loss": 0.4487,
+      "step": 1684
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.222533480865316e-06,
+      "loss": 0.4774,
+      "step": 1685
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.211239872124803e-06,
+      "loss": 0.4429,
+      "step": 1686
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.199951902906811e-06,
+      "loss": 0.406,
+      "step": 1687
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.188669590013336e-06,
+      "loss": 0.4742,
+      "step": 1688
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.177392950237963e-06,
+      "loss": 0.3943,
+      "step": 1689
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.166122000365835e-06,
+      "loss": 0.4689,
+      "step": 1690
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.154856757173615e-06,
+      "loss": 0.4358,
+      "step": 1691
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.143597237429481e-06,
+      "loss": 0.4677,
+      "step": 1692
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.132343457893093e-06,
+      "loss": 0.4137,
+      "step": 1693
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.121095435315554e-06,
+      "loss": 0.4925,
+      "step": 1694
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.109853186439411e-06,
+      "loss": 0.4333,
+      "step": 1695
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.098616727998612e-06,
+      "loss": 0.448,
+      "step": 1696
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.087386076718484e-06,
+      "loss": 0.4703,
+      "step": 1697
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.076161249315715e-06,
+      "loss": 0.4716,
+      "step": 1698
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.064942262498324e-06,
+      "loss": 0.4217,
+      "step": 1699
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.053729132965626e-06,
+      "loss": 0.4263,
+      "step": 1700
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0425218774082316e-06,
+      "loss": 0.4494,
+      "step": 1701
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.031320512508001e-06,
+      "loss": 0.4611,
+      "step": 1702
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.020125054938024e-06,
+      "loss": 0.4371,
+      "step": 1703
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.008935521362602e-06,
+      "loss": 0.4756,
+      "step": 1704
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.99775192843722e-06,
+      "loss": 0.4148,
+      "step": 1705
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.986574292808518e-06,
+      "loss": 0.4957,
+      "step": 1706
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.975402631114264e-06,
+      "loss": 0.4385,
+      "step": 1707
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.964236959983337e-06,
+      "loss": 0.4805,
+      "step": 1708
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.953077296035703e-06,
+      "loss": 0.4371,
+      "step": 1709
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9419236558823845e-06,
+      "loss": 0.4452,
+      "step": 1710
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.930776056125436e-06,
+      "loss": 0.4307,
+      "step": 1711
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.919634513357922e-06,
+      "loss": 0.442,
+      "step": 1712
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9084990441638905e-06,
+      "loss": 0.4182,
+      "step": 1713
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.897369665118347e-06,
+      "loss": 0.4149,
+      "step": 1714
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.886246392787235e-06,
+      "loss": 0.4314,
+      "step": 1715
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.87512924372741e-06,
+      "loss": 0.4711,
+      "step": 1716
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.864018234486611e-06,
+      "loss": 0.471,
+      "step": 1717
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.852913381603439e-06,
+      "loss": 0.4324,
+      "step": 1718
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.841814701607322e-06,
+      "loss": 0.4056,
+      "step": 1719
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.830722211018517e-06,
+      "loss": 0.4557,
+      "step": 1720
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8196359263480495e-06,
+      "loss": 0.502,
+      "step": 1721
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.808555864097727e-06,
+      "loss": 0.4549,
+      "step": 1722
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.797482040760074e-06,
+      "loss": 0.4189,
+      "step": 1723
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.786414472818352e-06,
+      "loss": 0.4201,
+      "step": 1724
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7753531767464895e-06,
+      "loss": 0.4061,
+      "step": 1725
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.764298169009093e-06,
+      "loss": 0.3972,
+      "step": 1726
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.753249466061404e-06,
+      "loss": 0.4224,
+      "step": 1727
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.742207084349274e-06,
+      "loss": 0.4698,
+      "step": 1728
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.731171040309159e-06,
+      "loss": 0.4561,
+      "step": 1729
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.720141350368072e-06,
+      "loss": 0.4348,
+      "step": 1730
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.70911803094357e-06,
+      "loss": 0.4998,
+      "step": 1731
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.698101098443724e-06,
+      "loss": 0.4899,
+      "step": 1732
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.687090569267102e-06,
+      "loss": 0.4427,
+      "step": 1733
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6760864598027465e-06,
+      "loss": 0.4298,
+      "step": 1734
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.665088786430129e-06,
+      "loss": 0.4263,
+      "step": 1735
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.65409756551916e-06,
+      "loss": 0.4249,
+      "step": 1736
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6431128134301335e-06,
+      "loss": 0.4308,
+      "step": 1737
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.632134546513706e-06,
+      "loss": 0.4133,
+      "step": 1738
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.621162781110905e-06,
+      "loss": 0.4631,
+      "step": 1739
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.610197533553058e-06,
+      "loss": 0.4021,
+      "step": 1740
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.599238820161807e-06,
+      "loss": 0.4316,
+      "step": 1741
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.58828665724906e-06,
+      "loss": 0.434,
+      "step": 1742
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.577341061116971e-06,
+      "loss": 0.4518,
+      "step": 1743
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.566402048057935e-06,
+      "loss": 0.4381,
+      "step": 1744
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.555469634354521e-06,
+      "loss": 0.4402,
+      "step": 1745
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.544543836279506e-06,
+      "loss": 0.3975,
+      "step": 1746
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.533624670095798e-06,
+      "loss": 0.4725,
+      "step": 1747
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.52271215205644e-06,
+      "loss": 0.4644,
+      "step": 1748
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.511806298404585e-06,
+      "loss": 0.4411,
+      "step": 1749
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.500907125373458e-06,
+      "loss": 0.4329,
+      "step": 1750
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.490014649186344e-06,
+      "loss": 0.44,
+      "step": 1751
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.479128886056553e-06,
+      "loss": 0.4165,
+      "step": 1752
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.468249852187418e-06,
+      "loss": 0.4126,
+      "step": 1753
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.457377563772241e-06,
+      "loss": 0.423,
+      "step": 1754
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.446512036994287e-06,
+      "loss": 0.4205,
+      "step": 1755
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.435653288026767e-06,
+      "loss": 0.4427,
+      "step": 1756
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.424801333032782e-06,
+      "loss": 0.4593,
+      "step": 1757
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.413956188165341e-06,
+      "loss": 0.434,
+      "step": 1758
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.403117869567305e-06,
+      "loss": 0.4606,
+      "step": 1759
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.392286393371373e-06,
+      "loss": 0.4189,
+      "step": 1760
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.381461775700072e-06,
+      "loss": 0.4724,
+      "step": 1761
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.370644032665706e-06,
+      "loss": 0.4285,
+      "step": 1762
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.359833180370353e-06,
+      "loss": 0.4788,
+      "step": 1763
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3490292349058305e-06,
+      "loss": 0.4403,
+      "step": 1764
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.338232212353676e-06,
+      "loss": 0.4614,
+      "step": 1765
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.32744212878513e-06,
+      "loss": 0.4729,
+      "step": 1766
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.316659000261089e-06,
+      "loss": 0.4399,
+      "step": 1767
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.305882842832119e-06,
+      "loss": 0.419,
+      "step": 1768
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.295113672538392e-06,
+      "loss": 0.4612,
+      "step": 1769
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.284351505409675e-06,
+      "loss": 0.4123,
+      "step": 1770
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.273596357465334e-06,
+      "loss": 0.4727,
+      "step": 1771
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2628482447142604e-06,
+      "loss": 0.3028,
+      "step": 1772
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2521071831549e-06,
+      "loss": 0.2416,
+      "step": 1773
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.241373188775183e-06,
+      "loss": 0.2963,
+      "step": 1774
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.230646277552528e-06,
+      "loss": 0.2924,
+      "step": 1775
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2199264654538084e-06,
+      "loss": 0.2475,
+      "step": 1776
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.209213768435331e-06,
+      "loss": 0.2766,
+      "step": 1777
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1985082024428155e-06,
+      "loss": 0.2256,
+      "step": 1778
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.187809783411366e-06,
+      "loss": 0.2572,
+      "step": 1779
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.177118527265438e-06,
+      "loss": 0.2348,
+      "step": 1780
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.166434449918848e-06,
+      "loss": 0.2373,
+      "step": 1781
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.155757567274696e-06,
+      "loss": 0.3081,
+      "step": 1782
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.145087895225402e-06,
+      "loss": 0.2569,
+      "step": 1783
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.134425449652634e-06,
+      "loss": 0.2218,
+      "step": 1784
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.123770246427315e-06,
+      "loss": 0.2377,
+      "step": 1785
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.113122301409581e-06,
+      "loss": 0.2328,
+      "step": 1786
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1024816304487635e-06,
+      "loss": 0.2539,
+      "step": 1787
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.091848249383379e-06,
+      "loss": 0.2515,
+      "step": 1788
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.081222174041068e-06,
+      "loss": 0.2458,
+      "step": 1789
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.070603420238625e-06,
+      "loss": 0.2615,
+      "step": 1790
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.05999200378193e-06,
+      "loss": 0.2637,
+      "step": 1791
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.049387940465939e-06,
+      "loss": 0.2165,
+      "step": 1792
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.038791246074677e-06,
+      "loss": 0.2414,
+      "step": 1793
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.028201936381189e-06,
+      "loss": 0.2265,
+      "step": 1794
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.017620027147534e-06,
+      "loss": 0.2307,
+      "step": 1795
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.007045534124748e-06,
+      "loss": 0.2201,
+      "step": 1796
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.996478473052832e-06,
+      "loss": 0.2179,
+      "step": 1797
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.985918859660732e-06,
+      "loss": 0.2492,
+      "step": 1798
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.975366709666295e-06,
+      "loss": 0.2683,
+      "step": 1799
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.964822038776277e-06,
+      "loss": 0.2277,
+      "step": 1800
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.954284862686277e-06,
+      "loss": 0.2802,
+      "step": 1801
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.943755197080753e-06,
+      "loss": 0.2312,
+      "step": 1802
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.933233057632989e-06,
+      "loss": 0.264,
+      "step": 1803
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.922718460005051e-06,
+      "loss": 0.2363,
+      "step": 1804
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.912211419847795e-06,
+      "loss": 0.2315,
+      "step": 1805
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9017119528008165e-06,
+      "loss": 0.2439,
+      "step": 1806
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.891220074492442e-06,
+      "loss": 0.2669,
+      "step": 1807
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.880735800539703e-06,
+      "loss": 0.2643,
+      "step": 1808
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.870259146548306e-06,
+      "loss": 0.2697,
+      "step": 1809
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.859790128112631e-06,
+      "loss": 0.2594,
+      "step": 1810
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.849328760815676e-06,
+      "loss": 0.2341,
+      "step": 1811
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.838875060229055e-06,
+      "loss": 0.216,
+      "step": 1812
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.828429041912981e-06,
+      "loss": 0.2614,
+      "step": 1813
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.817990721416208e-06,
+      "loss": 0.2237,
+      "step": 1814
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.807560114276059e-06,
+      "loss": 0.2688,
+      "step": 1815
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.797137236018356e-06,
+      "loss": 0.2223,
+      "step": 1816
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.786722102157431e-06,
+      "loss": 0.2498,
+      "step": 1817
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.77631472819608e-06,
+      "loss": 0.223,
+      "step": 1818
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.765915129625548e-06,
+      "loss": 0.2755,
+      "step": 1819
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.755523321925508e-06,
+      "loss": 0.2593,
+      "step": 1820
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.745139320564036e-06,
+      "loss": 0.2105,
+      "step": 1821
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.734763140997594e-06,
+      "loss": 0.2324,
+      "step": 1822
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.724394798670997e-06,
+      "loss": 0.217,
+      "step": 1823
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.714034309017385e-06,
+      "loss": 0.2558,
+      "step": 1824
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.703681687458231e-06,
+      "loss": 0.2129,
+      "step": 1825
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.693336949403278e-06,
+      "loss": 0.2689,
+      "step": 1826
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6830001102505384e-06,
+      "loss": 0.2311,
+      "step": 1827
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.672671185386273e-06,
+      "loss": 0.2445,
+      "step": 1828
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6623501901849534e-06,
+      "loss": 0.2254,
+      "step": 1829
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.652037140009259e-06,
+      "loss": 0.2451,
+      "step": 1830
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.641732050210032e-06,
+      "loss": 0.2514,
+      "step": 1831
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6314349361262815e-06,
+      "loss": 0.2324,
+      "step": 1832
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.621145813085117e-06,
+      "loss": 0.233,
+      "step": 1833
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.610864696401786e-06,
+      "loss": 0.2116,
+      "step": 1834
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.600591601379597e-06,
+      "loss": 0.2592,
+      "step": 1835
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.590326543309921e-06,
+      "loss": 0.2356,
+      "step": 1836
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.580069537472178e-06,
+      "loss": 0.2273,
+      "step": 1837
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.569820599133789e-06,
+      "loss": 0.2397,
+      "step": 1838
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.559579743550174e-06,
+      "loss": 0.2313,
+      "step": 1839
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.549346985964719e-06,
+      "loss": 0.2516,
+      "step": 1840
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5391223416087515e-06,
+      "loss": 0.2092,
+      "step": 1841
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5289058257015375e-06,
+      "loss": 0.2772,
+      "step": 1842
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.518697453450229e-06,
+      "loss": 0.2145,
+      "step": 1843
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.508497240049859e-06,
+      "loss": 0.2225,
+      "step": 1844
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.49830520068333e-06,
+      "loss": 0.2023,
+      "step": 1845
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.488121350521351e-06,
+      "loss": 0.2602,
+      "step": 1846
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.477945704722468e-06,
+      "loss": 0.2402,
+      "step": 1847
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.467778278432997e-06,
+      "loss": 0.2567,
+      "step": 1848
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.457619086787033e-06,
+      "loss": 0.2325,
+      "step": 1849
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.447468144906401e-06,
+      "loss": 0.2643,
+      "step": 1850
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.437325467900656e-06,
+      "loss": 0.2479,
+      "step": 1851
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4271910708670455e-06,
+      "loss": 0.2272,
+      "step": 1852
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4170649688904896e-06,
+      "loss": 0.223,
+      "step": 1853
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.406947177043574e-06,
+      "loss": 0.2434,
+      "step": 1854
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.396837710386503e-06,
+      "loss": 0.2548,
+      "step": 1855
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.386736583967087e-06,
+      "loss": 0.2318,
+      "step": 1856
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.37664381282074e-06,
+      "loss": 0.2105,
+      "step": 1857
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.366559411970413e-06,
+      "loss": 0.2558,
+      "step": 1858
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.356483396426621e-06,
+      "loss": 0.246,
+      "step": 1859
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.346415781187386e-06,
+      "loss": 0.247,
+      "step": 1860
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.336356581238223e-06,
+      "loss": 0.2399,
+      "step": 1861
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.326305811552136e-06,
+      "loss": 0.194,
+      "step": 1862
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.316263487089567e-06,
+      "loss": 0.2434,
+      "step": 1863
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.306229622798392e-06,
+      "loss": 0.2512,
+      "step": 1864
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.296204233613888e-06,
+      "loss": 0.2059,
+      "step": 1865
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.28618733445873e-06,
+      "loss": 0.2505,
+      "step": 1866
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.276178940242948e-06,
+      "loss": 0.2338,
+      "step": 1867
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2661790658639055e-06,
+      "loss": 0.2149,
+      "step": 1868
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.256187726206302e-06,
+      "loss": 0.25,
+      "step": 1869
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.246204936142116e-06,
+      "loss": 0.2085,
+      "step": 1870
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.23623071053061e-06,
+      "loss": 0.2244,
+      "step": 1871
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.226265064218293e-06,
+      "loss": 0.2217,
+      "step": 1872
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.216308012038903e-06,
+      "loss": 0.2456,
+      "step": 1873
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.206359568813397e-06,
+      "loss": 0.2139,
+      "step": 1874
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.196419749349905e-06,
+      "loss": 0.2453,
+      "step": 1875
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1864885684437255e-06,
+      "loss": 0.2332,
+      "step": 1876
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.176566040877297e-06,
+      "loss": 0.2641,
+      "step": 1877
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.166652181420177e-06,
+      "loss": 0.2431,
+      "step": 1878
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1567470048290295e-06,
+      "loss": 0.2306,
+      "step": 1879
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1468505258475785e-06,
+      "loss": 0.2325,
+      "step": 1880
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.136962759206619e-06,
+      "loss": 0.2249,
+      "step": 1881
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.127083719623969e-06,
+      "loss": 0.2507,
+      "step": 1882
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.117213421804445e-06,
+      "loss": 0.2238,
+      "step": 1883
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.107351880439875e-06,
+      "loss": 0.2618,
+      "step": 1884
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.097499110209032e-06,
+      "loss": 0.2426,
+      "step": 1885
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.087655125777652e-06,
+      "loss": 0.226,
+      "step": 1886
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.077819941798379e-06,
+      "loss": 0.2413,
+      "step": 1887
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.067993572910759e-06,
+      "loss": 0.2361,
+      "step": 1888
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.058176033741233e-06,
+      "loss": 0.2581,
+      "step": 1889
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0483673389030675e-06,
+      "loss": 0.255,
+      "step": 1890
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.038567502996398e-06,
+      "loss": 0.2622,
+      "step": 1891
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.028776540608155e-06,
+      "loss": 0.2381,
+      "step": 1892
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.01899446631206e-06,
+      "loss": 0.2327,
+      "step": 1893
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.009221294668615e-06,
+      "loss": 0.2532,
+      "step": 1894
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9994570402250656e-06,
+      "loss": 0.2414,
+      "step": 1895
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.989701717515379e-06,
+      "loss": 0.2415,
+      "step": 1896
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.979955341060233e-06,
+      "loss": 0.2227,
+      "step": 1897
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9702179253669925e-06,
+      "loss": 0.272,
+      "step": 1898
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.960489484929677e-06,
+      "loss": 0.2416,
+      "step": 1899
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.950770034228946e-06,
+      "loss": 0.2389,
+      "step": 1900
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9410595877320934e-06,
+      "loss": 0.2132,
+      "step": 1901
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.931358159892982e-06,
+      "loss": 0.2462,
+      "step": 1902
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.921665765152079e-06,
+      "loss": 0.2143,
+      "step": 1903
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.911982417936392e-06,
+      "loss": 0.2417,
+      "step": 1904
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.902308132659457e-06,
+      "loss": 0.2527,
+      "step": 1905
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.892642923721337e-06,
+      "loss": 0.2082,
+      "step": 1906
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.882986805508571e-06,
+      "loss": 0.2236,
+      "step": 1907
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.87333979239417e-06,
+      "loss": 0.2094,
+      "step": 1908
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.863701898737596e-06,
+      "loss": 0.2609,
+      "step": 1909
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.854073138884731e-06,
+      "loss": 0.2386,
+      "step": 1910
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.84445352716787e-06,
+      "loss": 0.23,
+      "step": 1911
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.83484307790568e-06,
+      "loss": 0.2479,
+      "step": 1912
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.825241805403201e-06,
+      "loss": 0.2247,
+      "step": 1913
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.815649723951808e-06,
+      "loss": 0.2263,
+      "step": 1914
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.806066847829192e-06,
+      "loss": 0.2548,
+      "step": 1915
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7964931912993473e-06,
+      "loss": 0.2167,
+      "step": 1916
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.786928768612539e-06,
+      "loss": 0.2596,
+      "step": 1917
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.777373594005298e-06,
+      "loss": 0.2029,
+      "step": 1918
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7678276817003824e-06,
+      "loss": 0.2658,
+      "step": 1919
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.758291045906761e-06,
+      "loss": 0.2347,
+      "step": 1920
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.748763700819601e-06,
+      "loss": 0.2448,
+      "step": 1921
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.739245660620232e-06,
+      "loss": 0.2147,
+      "step": 1922
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.729736939476147e-06,
+      "loss": 0.2481,
+      "step": 1923
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7202375515409562e-06,
+      "loss": 0.2092,
+      "step": 1924
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7107475109543767e-06,
+      "loss": 0.2632,
+      "step": 1925
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7012668318422227e-06,
+      "loss": 0.2371,
+      "step": 1926
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6917955283163666e-06,
+      "loss": 0.2596,
+      "step": 1927
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6823336144747248e-06,
+      "loss": 0.2518,
+      "step": 1928
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6728811044012345e-06,
+      "loss": 0.2267,
+      "step": 1929
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6634380121658484e-06,
+      "loss": 0.2105,
+      "step": 1930
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.654004351824488e-06,
+      "loss": 0.2416,
+      "step": 1931
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.644580137419037e-06,
+      "loss": 0.2704,
+      "step": 1932
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6351653829773315e-06,
+      "loss": 0.2124,
+      "step": 1933
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.625760102513103e-06,
+      "loss": 0.2485,
+      "step": 1934
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.616364310026006e-06,
+      "loss": 0.2129,
+      "step": 1935
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.606978019501557e-06,
+      "loss": 0.2389,
+      "step": 1936
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.597601244911131e-06,
+      "loss": 0.2126,
+      "step": 1937
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5882340002119466e-06,
+      "loss": 0.2147,
+      "step": 1938
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5788762993470284e-06,
+      "loss": 0.2589,
+      "step": 1939
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5695281562451965e-06,
+      "loss": 0.206,
+      "step": 1940
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.56018958482105e-06,
+      "loss": 0.2062,
+      "step": 1941
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.55086059897493e-06,
+      "loss": 0.215,
+      "step": 1942
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.541541212592924e-06,
+      "loss": 0.2246,
+      "step": 1943
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.532231439546817e-06,
+      "loss": 0.2518,
+      "step": 1944
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5229312936941017e-06,
+      "loss": 0.2861,
+      "step": 1945
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5136407888779143e-06,
+      "loss": 0.2574,
+      "step": 1946
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.504359938927068e-06,
+      "loss": 0.2216,
+      "step": 1947
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.495088757655989e-06,
+      "loss": 0.2364,
+      "step": 1948
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4858272588647124e-06,
+      "loss": 0.2523,
+      "step": 1949
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.476575456338871e-06,
+      "loss": 0.238,
+      "step": 1950
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.467333363849655e-06,
+      "loss": 0.2504,
+      "step": 1951
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.458100995153806e-06,
+      "loss": 0.2251,
+      "step": 1952
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4488783639935875e-06,
+      "loss": 0.2495,
+      "step": 1953
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.43966548409677e-06,
+      "loss": 0.2223,
+      "step": 1954
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4304623691766193e-06,
+      "loss": 0.216,
+      "step": 1955
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.421269032931852e-06,
+      "loss": 0.2234,
+      "step": 1956
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.412085489046634e-06,
+      "loss": 0.2357,
+      "step": 1957
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.402911751190565e-06,
+      "loss": 0.2212,
+      "step": 1958
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3937478330186267e-06,
+      "loss": 0.2181,
+      "step": 1959
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.38459374817121e-06,
+      "loss": 0.2213,
+      "step": 1960
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.375449510274048e-06,
+      "loss": 0.2458,
+      "step": 1961
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3663151329382327e-06,
+      "loss": 0.2242,
+      "step": 1962
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3571906297601697e-06,
+      "loss": 0.2207,
+      "step": 1963
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.348076014321566e-06,
+      "loss": 0.232,
+      "step": 1964
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3389713001894163e-06,
+      "loss": 0.2268,
+      "step": 1965
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3298765009159704e-06,
+      "loss": 0.2102,
+      "step": 1966
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3207916300387322e-06,
+      "loss": 0.2107,
+      "step": 1967
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3117167010804142e-06,
+      "loss": 0.2407,
+      "step": 1968
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3026517275489346e-06,
+      "loss": 0.2152,
+      "step": 1969
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.293596722937399e-06,
+      "loss": 0.2598,
+      "step": 1970
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2845517007240666e-06,
+      "loss": 0.2322,
+      "step": 1971
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.275516674372341e-06,
+      "loss": 0.2295,
+      "step": 1972
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2664916573307483e-06,
+      "loss": 0.2598,
+      "step": 1973
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.25747666303291e-06,
+      "loss": 0.2444,
+      "step": 1974
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.24847170489754e-06,
+      "loss": 0.2552,
+      "step": 1975
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.239476796328401e-06,
+      "loss": 0.2339,
+      "step": 1976
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.230491950714312e-06,
+      "loss": 0.2164,
+      "step": 1977
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2215171814290924e-06,
+      "loss": 0.2577,
+      "step": 1978
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.212552501831584e-06,
+      "loss": 0.2545,
+      "step": 1979
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.203597925265598e-06,
+      "loss": 0.2496,
+      "step": 1980
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1946534650599083e-06,
+      "loss": 0.2263,
+      "step": 1981
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.185719134528239e-06,
+      "loss": 0.2489,
+      "step": 1982
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.176794946969227e-06,
+      "loss": 0.2588,
+      "step": 1983
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.167880915666416e-06,
+      "loss": 0.2252,
+      "step": 1984
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1589770538882303e-06,
+      "loss": 0.2158,
+      "step": 1985
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1500833748879545e-06,
+      "loss": 0.1977,
+      "step": 1986
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1411998919037268e-06,
+      "loss": 0.2204,
+      "step": 1987
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1323266181584967e-06,
+      "loss": 0.218,
+      "step": 1988
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.123463566860021e-06,
+      "loss": 0.2326,
+      "step": 1989
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1146107512008505e-06,
+      "loss": 0.2027,
+      "step": 1990
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1057681843582786e-06,
+      "loss": 0.2447,
+      "step": 1991
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.096935879494367e-06,
+      "loss": 0.2337,
+      "step": 1992
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.088113849755885e-06,
+      "loss": 0.2758,
+      "step": 1993
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0793021082743213e-06,
+      "loss": 0.2472,
+      "step": 1994
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.07050066816584e-06,
+      "loss": 0.2366,
+      "step": 1995
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0617095425312772e-06,
+      "loss": 0.2436,
+      "step": 1996
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.052928744456113e-06,
+      "loss": 0.2017,
+      "step": 1997
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0441582870104537e-06,
+      "loss": 0.2374,
+      "step": 1998
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.035398183249023e-06,
+      "loss": 0.2419,
+      "step": 1999
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0266484462111244e-06,
+      "loss": 0.2449,
+      "step": 2000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0179090889206285e-06,
+      "loss": 0.2384,
+      "step": 2001
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.009180124385972e-06,
+      "loss": 0.2194,
+      "step": 2002
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.000461565600096e-06,
+      "loss": 0.2295,
+      "step": 2003
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9917534255404767e-06,
+      "loss": 0.2313,
+      "step": 2004
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.98305571716907e-06,
+      "loss": 0.2267,
+      "step": 2005
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.974368453432305e-06,
+      "loss": 0.2469,
+      "step": 2006
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.965691647261071e-06,
+      "loss": 0.2355,
+      "step": 2007
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9570253115706802e-06,
+      "loss": 0.2216,
+      "step": 2008
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.948369459260877e-06,
+      "loss": 0.2387,
+      "step": 2009
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9397241032157764e-06,
+      "loss": 0.2594,
+      "step": 2010
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9310892563038927e-06,
+      "loss": 0.2347,
+      "step": 2011
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9224649313780875e-06,
+      "loss": 0.2514,
+      "step": 2012
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9138511412755553e-06,
+      "loss": 0.2366,
+      "step": 2013
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9052478988178256e-06,
+      "loss": 0.2565,
+      "step": 2014
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8966552168107133e-06,
+      "loss": 0.2643,
+      "step": 2015
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8880731080443182e-06,
+      "loss": 0.2441,
+      "step": 2016
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8795015852930055e-06,
+      "loss": 0.2517,
+      "step": 2017
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8709406613153757e-06,
+      "loss": 0.2184,
+      "step": 2018
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.862390348854265e-06,
+      "loss": 0.2457,
+      "step": 2019
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8538506606367033e-06,
+      "loss": 0.2485,
+      "step": 2020
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.845321609373912e-06,
+      "loss": 0.2263,
+      "step": 2021
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8368032077612785e-06,
+      "loss": 0.2249,
+      "step": 2022
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8282954684783337e-06,
+      "loss": 0.2458,
+      "step": 2023
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8197984041887492e-06,
+      "loss": 0.233,
+      "step": 2024
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8113120275402937e-06,
+      "loss": 0.2586,
+      "step": 2025
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8028363511648393e-06,
+      "loss": 0.2586,
+      "step": 2026
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.794371387678323e-06,
+      "loss": 0.2468,
+      "step": 2027
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.785917149680738e-06,
+      "loss": 0.2406,
+      "step": 2028
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.777473649756113e-06,
+      "loss": 0.2316,
+      "step": 2029
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7690409004724883e-06,
+      "loss": 0.2006,
+      "step": 2030
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.760618914381915e-06,
+      "loss": 0.226,
+      "step": 2031
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.752207704020411e-06,
+      "loss": 0.2731,
+      "step": 2032
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7438072819079553e-06,
+      "loss": 0.2502,
+      "step": 2033
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7354176605484816e-06,
+      "loss": 0.2367,
+      "step": 2034
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7270388524298262e-06,
+      "loss": 0.2198,
+      "step": 2035
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7186708700237494e-06,
+      "loss": 0.2664,
+      "step": 2036
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7103137257858867e-06,
+      "loss": 0.2409,
+      "step": 2037
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.70196743215574e-06,
+      "loss": 0.2082,
+      "step": 2038
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6936320015566708e-06,
+      "loss": 0.2447,
+      "step": 2039
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6853074463958618e-06,
+      "loss": 0.2131,
+      "step": 2040
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6769937790643084e-06,
+      "loss": 0.2219,
+      "step": 2041
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6686910119367995e-06,
+      "loss": 0.2083,
+      "step": 2042
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.660399157371907e-06,
+      "loss": 0.2299,
+      "step": 2043
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6521182277119517e-06,
+      "loss": 0.2305,
+      "step": 2044
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6438482352829896e-06,
+      "loss": 0.2443,
+      "step": 2045
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.635589192394813e-06,
+      "loss": 0.2109,
+      "step": 2046
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6273411113408918e-06,
+      "loss": 0.2194,
+      "step": 2047
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.619104004398403e-06,
+      "loss": 0.2408,
+      "step": 2048
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6108778838281735e-06,
+      "loss": 0.2518,
+      "step": 2049
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6026627618746793e-06,
+      "loss": 0.242,
+      "step": 2050
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.594458650766034e-06,
+      "loss": 0.2231,
+      "step": 2051
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5862655627139534e-06,
+      "loss": 0.2096,
+      "step": 2052
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5780835099137446e-06,
+      "loss": 0.2216,
+      "step": 2053
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.569912504544295e-06,
+      "loss": 0.2273,
+      "step": 2054
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.56175255876804e-06,
+      "loss": 0.2178,
+      "step": 2055
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5536036847309643e-06,
+      "loss": 0.2499,
+      "step": 2056
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.54546589456256e-06,
+      "loss": 0.2548,
+      "step": 2057
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5373392003758333e-06,
+      "loss": 0.2461,
+      "step": 2058
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5292236142672645e-06,
+      "loss": 0.2355,
+      "step": 2059
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.521119148316803e-06,
+      "loss": 0.2151,
+      "step": 2060
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.513025814587845e-06,
+      "loss": 0.255,
+      "step": 2061
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5049436251272164e-06,
+      "loss": 0.2247,
+      "step": 2062
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4968725919651614e-06,
+      "loss": 0.1983,
+      "step": 2063
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4888127271153095e-06,
+      "loss": 0.2283,
+      "step": 2064
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4807640425746693e-06,
+      "loss": 0.2245,
+      "step": 2065
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4727265503236085e-06,
+      "loss": 0.2533,
+      "step": 2066
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.464700262325833e-06,
+      "loss": 0.2077,
+      "step": 2067
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4566851905283774e-06,
+      "loss": 0.2418,
+      "step": 2068
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4486813468615734e-06,
+      "loss": 0.2423,
+      "step": 2069
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4406887432390426e-06,
+      "loss": 0.2213,
+      "step": 2070
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.432707391557679e-06,
+      "loss": 0.2406,
+      "step": 2071
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4247373036976262e-06,
+      "loss": 0.2351,
+      "step": 2072
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4167784915222592e-06,
+      "loss": 0.2403,
+      "step": 2073
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4088309668781686e-06,
+      "loss": 0.2051,
+      "step": 2074
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.400894741595152e-06,
+      "loss": 0.2138,
+      "step": 2075
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3929698274861814e-06,
+      "loss": 0.1989,
+      "step": 2076
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3850562363473884e-06,
+      "loss": 0.2273,
+      "step": 2077
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3771539799580645e-06,
+      "loss": 0.2158,
+      "step": 2078
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3692630700806096e-06,
+      "loss": 0.2488,
+      "step": 2079
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3613835184605527e-06,
+      "loss": 0.2536,
+      "step": 2080
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.353515336826506e-06,
+      "loss": 0.2435,
+      "step": 2081
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.34565853689016e-06,
+      "loss": 0.2386,
+      "step": 2082
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.337813130346267e-06,
+      "loss": 0.2304,
+      "step": 2083
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.329979128872615e-06,
+      "loss": 0.2027,
+      "step": 2084
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.32215654413002e-06,
+      "loss": 0.2396,
+      "step": 2085
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3143453877623e-06,
+      "loss": 0.2563,
+      "step": 2086
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3065456713962632e-06,
+      "loss": 0.2268,
+      "step": 2087
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2987574066416963e-06,
+      "loss": 0.2573,
+      "step": 2088
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2909806050913284e-06,
+      "loss": 0.1902,
+      "step": 2089
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2832152783208393e-06,
+      "loss": 0.2145,
+      "step": 2090
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2754614378888186e-06,
+      "loss": 0.2125,
+      "step": 2091
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.267719095336761e-06,
+      "loss": 0.2827,
+      "step": 2092
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2599882621890467e-06,
+      "loss": 0.2512,
+      "step": 2093
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2522689499529247e-06,
+      "loss": 0.2364,
+      "step": 2094
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2445611701185e-06,
+      "loss": 0.2334,
+      "step": 2095
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2368649341587033e-06,
+      "loss": 0.2278,
+      "step": 2096
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2291802535292885e-06,
+      "loss": 0.2197,
+      "step": 2097
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2215071396688058e-06,
+      "loss": 0.2638,
+      "step": 2098
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2138456039985902e-06,
+      "loss": 0.2288,
+      "step": 2099
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.206195657922745e-06,
+      "loss": 0.2317,
+      "step": 2100
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.198557312828118e-06,
+      "loss": 0.2381,
+      "step": 2101
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1909305800842904e-06,
+      "loss": 0.2228,
+      "step": 2102
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1833154710435657e-06,
+      "loss": 0.2337,
+      "step": 2103
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.175711997040929e-06,
+      "loss": 0.2307,
+      "step": 2104
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1681201693940667e-06,
+      "loss": 0.2382,
+      "step": 2105
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.160539999403313e-06,
+      "loss": 0.2147,
+      "step": 2106
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1529714983516624e-06,
+      "loss": 0.29,
+      "step": 2107
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1454146775047334e-06,
+      "loss": 0.2251,
+      "step": 2108
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.137869548110757e-06,
+      "loss": 0.1961,
+      "step": 2109
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1303361214005723e-06,
+      "loss": 0.2607,
+      "step": 2110
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1228144085875825e-06,
+      "loss": 0.2293,
+      "step": 2111
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1153044208677707e-06,
+      "loss": 0.247,
+      "step": 2112
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1078061694196584e-06,
+      "loss": 0.238,
+      "step": 2113
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1003196654042966e-06,
+      "loss": 0.232,
+      "step": 2114
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0928449199652602e-06,
+      "loss": 0.2264,
+      "step": 2115
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0853819442286118e-06,
+      "loss": 0.2576,
+      "step": 2116
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.077930749302898e-06,
+      "loss": 0.2609,
+      "step": 2117
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.070491346279131e-06,
+      "loss": 0.2411,
+      "step": 2118
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.063063746230768e-06,
+      "loss": 0.2208,
+      "step": 2119
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0556479602137036e-06,
+      "loss": 0.2319,
+      "step": 2120
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0482439992662375e-06,
+      "loss": 0.2437,
+      "step": 2121
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0408518744090842e-06,
+      "loss": 0.2543,
+      "step": 2122
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.033471596645318e-06,
+      "loss": 0.1898,
+      "step": 2123
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0261031769603978e-06,
+      "loss": 0.2244,
+      "step": 2124
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0187466263221243e-06,
+      "loss": 0.2387,
+      "step": 2125
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0114019556806286e-06,
+      "loss": 0.234,
+      "step": 2126
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.004069175968367e-06,
+      "loss": 0.2177,
+      "step": 2127
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9967482981000896e-06,
+      "loss": 0.2374,
+      "step": 2128
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.98943933297283e-06,
+      "loss": 0.2095,
+      "step": 2129
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.982142291465896e-06,
+      "loss": 0.2533,
+      "step": 2130
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.974857184440837e-06,
+      "loss": 0.1914,
+      "step": 2131
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.967584022741453e-06,
+      "loss": 0.2007,
+      "step": 2132
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9603228171937505e-06,
+      "loss": 0.2479,
+      "step": 2133
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.953073578605942e-06,
+      "loss": 0.2848,
+      "step": 2134
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9458363177684368e-06,
+      "loss": 0.2343,
+      "step": 2135
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.938611045453799e-06,
+      "loss": 0.27,
+      "step": 2136
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.931397772416762e-06,
+      "loss": 0.1795,
+      "step": 2137
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9241965093941906e-06,
+      "loss": 0.2186,
+      "step": 2138
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.917007267105081e-06,
+      "loss": 0.2423,
+      "step": 2139
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 0.2364,
+      "step": 2140
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9026648875137177e-06,
+      "loss": 0.2595,
+      "step": 2141
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8955117715599203e-06,
+      "loss": 0.2078,
+      "step": 2142
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8883707190364552e-06,
+      "loss": 0.2357,
+      "step": 2143
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8812417405726969e-06,
+      "loss": 0.2456,
+      "step": 2144
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8741248467800366e-06,
+      "loss": 0.2386,
+      "step": 2145
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.867020048251883e-06,
+      "loss": 0.2156,
+      "step": 2146
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8599273555636466e-06,
+      "loss": 0.2526,
+      "step": 2147
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8528467792727023e-06,
+      "loss": 0.2173,
+      "step": 2148
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8457783299184074e-06,
+      "loss": 0.2393,
+      "step": 2149
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.838722018022061e-06,
+      "loss": 0.2812,
+      "step": 2150
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8316778540868907e-06,
+      "loss": 0.2187,
+      "step": 2151
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8246458485980545e-06,
+      "loss": 0.2553,
+      "step": 2152
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8176260120225985e-06,
+      "loss": 0.2173,
+      "step": 2153
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8106183548094725e-06,
+      "loss": 0.2093,
+      "step": 2154
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8036228873894745e-06,
+      "loss": 0.2183,
+      "step": 2155
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7966396201752812e-06,
+      "loss": 0.2236,
+      "step": 2156
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7896685635613942e-06,
+      "loss": 0.2249,
+      "step": 2157
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7827097279241446e-06,
+      "loss": 0.2336,
+      "step": 2158
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7757631236216777e-06,
+      "loss": 0.2286,
+      "step": 2159
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7688287609939248e-06,
+      "loss": 0.2272,
+      "step": 2160
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7619066503625991e-06,
+      "loss": 0.2139,
+      "step": 2161
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7549968020311759e-06,
+      "loss": 0.246,
+      "step": 2162
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7480992262848773e-06,
+      "loss": 0.2604,
+      "step": 2163
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.741213933390664e-06,
+      "loss": 0.2201,
+      "step": 2164
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7343409335972071e-06,
+      "loss": 0.2067,
+      "step": 2165
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7274802371348809e-06,
+      "loss": 0.2076,
+      "step": 2166
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7206318542157474e-06,
+      "loss": 0.2502,
+      "step": 2167
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.713795795033537e-06,
+      "loss": 0.2181,
+      "step": 2168
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.706972069763645e-06,
+      "loss": 0.2226,
+      "step": 2169
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.700160688563095e-06,
+      "loss": 0.2362,
+      "step": 2170
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6933616615705505e-06,
+      "loss": 0.2451,
+      "step": 2171
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6865749989062764e-06,
+      "loss": 0.2318,
+      "step": 2172
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6798007106721349e-06,
+      "loss": 0.2378,
+      "step": 2173
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6730388069515713e-06,
+      "loss": 0.2588,
+      "step": 2174
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6662892978095912e-06,
+      "loss": 0.2407,
+      "step": 2175
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6595521932927628e-06,
+      "loss": 0.2221,
+      "step": 2176
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6528275034291774e-06,
+      "loss": 0.2506,
+      "step": 2177
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6461152382284528e-06,
+      "loss": 0.2056,
+      "step": 2178
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6394154076817182e-06,
+      "loss": 0.221,
+      "step": 2179
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6327280217615793e-06,
+      "loss": 0.2144,
+      "step": 2180
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.626053090422135e-06,
+      "loss": 0.2105,
+      "step": 2181
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6193906235989343e-06,
+      "loss": 0.2599,
+      "step": 2182
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6127406312089755e-06,
+      "loss": 0.2125,
+      "step": 2183
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6061031231506928e-06,
+      "loss": 0.2361,
+      "step": 2184
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5994781093039336e-06,
+      "loss": 0.243,
+      "step": 2185
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5928655995299492e-06,
+      "loss": 0.231,
+      "step": 2186
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5862656036713753e-06,
+      "loss": 0.2352,
+      "step": 2187
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5796781315522302e-06,
+      "loss": 0.22,
+      "step": 2188
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5731031929778796e-06,
+      "loss": 0.2128,
+      "step": 2189
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.566540797735039e-06,
+      "loss": 0.2487,
+      "step": 2190
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5599909555917535e-06,
+      "loss": 0.2213,
+      "step": 2191
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5534536762973828e-06,
+      "loss": 0.2324,
+      "step": 2192
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.546928969582584e-06,
+      "loss": 0.2262,
+      "step": 2193
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5404168451593037e-06,
+      "loss": 0.2332,
+      "step": 2194
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5339173127207564e-06,
+      "loss": 0.2242,
+      "step": 2195
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5274303819414182e-06,
+      "loss": 0.2221,
+      "step": 2196
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5209560624770015e-06,
+      "loss": 0.2398,
+      "step": 2197
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5144943639644582e-06,
+      "loss": 0.2024,
+      "step": 2198
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5080452960219405e-06,
+      "loss": 0.2104,
+      "step": 2199
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5016088682488027e-06,
+      "loss": 0.1969,
+      "step": 2200
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.495185090225596e-06,
+      "loss": 0.2405,
+      "step": 2201
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.488773971514028e-06,
+      "loss": 0.2349,
+      "step": 2202
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4823755216569747e-06,
+      "loss": 0.2448,
+      "step": 2203
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.475989750178446e-06,
+      "loss": 0.2276,
+      "step": 2204
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4696166665835853e-06,
+      "loss": 0.2236,
+      "step": 2205
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.463256280358648e-06,
+      "loss": 0.2477,
+      "step": 2206
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4569086009709876e-06,
+      "loss": 0.2414,
+      "step": 2207
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4505736378690504e-06,
+      "loss": 0.2282,
+      "step": 2208
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4442514004823495e-06,
+      "loss": 0.194,
+      "step": 2209
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4379418982214544e-06,
+      "loss": 0.2407,
+      "step": 2210
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4316451404779862e-06,
+      "loss": 0.2277,
+      "step": 2211
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4253611366245835e-06,
+      "loss": 0.2357,
+      "step": 2212
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4190898960149146e-06,
+      "loss": 0.2331,
+      "step": 2213
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4128314279836407e-06,
+      "loss": 0.2331,
+      "step": 2214
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4065857418464123e-06,
+      "loss": 0.2354,
+      "step": 2215
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4003528468998594e-06,
+      "loss": 0.2488,
+      "step": 2216
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3941327524215686e-06,
+      "loss": 0.2507,
+      "step": 2217
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3879254676700715e-06,
+      "loss": 0.2457,
+      "step": 2218
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3817310018848328e-06,
+      "loss": 0.2382,
+      "step": 2219
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.375549364286244e-06,
+      "loss": 0.2312,
+      "step": 2220
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3693805640755909e-06,
+      "loss": 0.2395,
+      "step": 2221
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3632246104350567e-06,
+      "loss": 0.2158,
+      "step": 2222
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.357081512527708e-06,
+      "loss": 0.2036,
+      "step": 2223
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.350951279497461e-06,
+      "loss": 0.2117,
+      "step": 2224
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3448339204690975e-06,
+      "loss": 0.2229,
+      "step": 2225
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.33872944454823e-06,
+      "loss": 0.2166,
+      "step": 2226
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3326378608212908e-06,
+      "loss": 0.2379,
+      "step": 2227
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3265591783555343e-06,
+      "loss": 0.2253,
+      "step": 2228
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3204934061989993e-06,
+      "loss": 0.2559,
+      "step": 2229
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3144405533805138e-06,
+      "loss": 0.2265,
+      "step": 2230
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.308400628909673e-06,
+      "loss": 0.219,
+      "step": 2231
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3023736417768285e-06,
+      "loss": 0.2496,
+      "step": 2232
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.296359600953081e-06,
+      "loss": 0.2485,
+      "step": 2233
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2903585153902498e-06,
+      "loss": 0.2719,
+      "step": 2234
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2843703940208818e-06,
+      "loss": 0.2255,
+      "step": 2235
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2783952457582193e-06,
+      "loss": 0.2406,
+      "step": 2236
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2724330794961958e-06,
+      "loss": 0.2081,
+      "step": 2237
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2664839041094224e-06,
+      "loss": 0.2468,
+      "step": 2238
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2605477284531708e-06,
+      "loss": 0.2259,
+      "step": 2239
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.254624561363369e-06,
+      "loss": 0.2257,
+      "step": 2240
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2487144116565764e-06,
+      "loss": 0.2413,
+      "step": 2241
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2428172881299772e-06,
+      "loss": 0.2361,
+      "step": 2242
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2369331995613664e-06,
+      "loss": 0.2655,
+      "step": 2243
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.231062154709135e-06,
+      "loss": 0.2188,
+      "step": 2244
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2252041623122646e-06,
+      "loss": 0.232,
+      "step": 2245
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2193592310903023e-06,
+      "loss": 0.2334,
+      "step": 2246
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2135273697433535e-06,
+      "loss": 0.2222,
+      "step": 2247
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.20770858695208e-06,
+      "loss": 0.1969,
+      "step": 2248
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2019028913776565e-06,
+      "loss": 0.2391,
+      "step": 2249
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1961102916617962e-06,
+      "loss": 0.2364,
+      "step": 2250
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1903307964267085e-06,
+      "loss": 0.2307,
+      "step": 2251
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1845644142751046e-06,
+      "loss": 0.2514,
+      "step": 2252
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1788111537901703e-06,
+      "loss": 0.2438,
+      "step": 2253
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1730710235355602e-06,
+      "loss": 0.2462,
+      "step": 2254
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1673440320553941e-06,
+      "loss": 0.2156,
+      "step": 2255
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1616301878742186e-06,
+      "loss": 0.2092,
+      "step": 2256
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1559294994970283e-06,
+      "loss": 0.2336,
+      "step": 2257
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.150241975409222e-06,
+      "loss": 0.224,
+      "step": 2258
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1445676240766102e-06,
+      "loss": 0.223,
+      "step": 2259
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1389064539453953e-06,
+      "loss": 0.2082,
+      "step": 2260
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1332584734421603e-06,
+      "loss": 0.2209,
+      "step": 2261
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1276236909738513e-06,
+      "loss": 0.2411,
+      "step": 2262
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1220021149277739e-06,
+      "loss": 0.2168,
+      "step": 2263
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1163937536715708e-06,
+      "loss": 0.2363,
+      "step": 2264
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1107986155532247e-06,
+      "loss": 0.2538,
+      "step": 2265
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1052167089010235e-06,
+      "loss": 0.2149,
+      "step": 2266
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0996480420235723e-06,
+      "loss": 0.2261,
+      "step": 2267
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0940926232097549e-06,
+      "loss": 0.2135,
+      "step": 2268
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.08855046072875e-06,
+      "loss": 0.2449,
+      "step": 2269
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0830215628299956e-06,
+      "loss": 0.2269,
+      "step": 2270
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0775059377431851e-06,
+      "loss": 0.244,
+      "step": 2271
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0720035936782626e-06,
+      "loss": 0.226,
+      "step": 2272
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0665145388253973e-06,
+      "loss": 0.2267,
+      "step": 2273
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0610387813549782e-06,
+      "loss": 0.1951,
+      "step": 2274
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0555763294176047e-06,
+      "loss": 0.2024,
+      "step": 2275
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0501271911440646e-06,
+      "loss": 0.2194,
+      "step": 2276
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.044691374645338e-06,
+      "loss": 0.2302,
+      "step": 2277
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0392688880125657e-06,
+      "loss": 0.2538,
+      "step": 2278
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0338597393170558e-06,
+      "loss": 0.2045,
+      "step": 2279
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.02846393661026e-06,
+      "loss": 0.2205,
+      "step": 2280
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0230814879237573e-06,
+      "loss": 0.2197,
+      "step": 2281
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0177124012692618e-06,
+      "loss": 0.2582,
+      "step": 2282
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.012356684638589e-06,
+      "loss": 0.246,
+      "step": 2283
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0070143460036596e-06,
+      "loss": 0.2379,
+      "step": 2284
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0016853933164773e-06,
+      "loss": 0.2164,
+      "step": 2285
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.963698345091222e-07,
+      "loss": 0.2479,
+      "step": 2286
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.910676774937377e-07,
+      "loss": 0.2153,
+      "step": 2287
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.857789301625176e-07,
+      "loss": 0.2435,
+      "step": 2288
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.80503600387699e-07,
+      "loss": 0.2719,
+      "step": 2289
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.75241696021544e-07,
+      "loss": 0.2533,
+      "step": 2290
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.699932248963307e-07,
+      "loss": 0.2624,
+      "step": 2291
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.647581948243456e-07,
+      "loss": 0.2044,
+      "step": 2292
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.595366135978657e-07,
+      "loss": 0.2409,
+      "step": 2293
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.54328488989149e-07,
+      "loss": 0.2291,
+      "step": 2294
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.491338287504249e-07,
+      "loss": 0.2284,
+      "step": 2295
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.439526406138788e-07,
+      "loss": 0.1996,
+      "step": 2296
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.387849322916487e-07,
+      "loss": 0.251,
+      "step": 2297
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.336307114758014e-07,
+      "loss": 0.209,
+      "step": 2298
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.284899858383356e-07,
+      "loss": 0.2344,
+      "step": 2299
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.233627630311503e-07,
+      "loss": 0.2525,
+      "step": 2300
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.182490506860597e-07,
+      "loss": 0.21,
+      "step": 2301
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.131488564147583e-07,
+      "loss": 0.2355,
+      "step": 2302
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.080621878088203e-07,
+      "loss": 0.2184,
+      "step": 2303
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.029890524396934e-07,
+      "loss": 0.2144,
+      "step": 2304
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.979294578586739e-07,
+      "loss": 0.253,
+      "step": 2305
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.928834115969065e-07,
+      "loss": 0.2092,
+      "step": 2306
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.878509211653675e-07,
+      "loss": 0.2029,
+      "step": 2307
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.828319940548557e-07,
+      "loss": 0.2447,
+      "step": 2308
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.778266377359856e-07,
+      "loss": 0.2757,
+      "step": 2309
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.728348596591641e-07,
+      "loss": 0.2099,
+      "step": 2310
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.678566672545952e-07,
+      "loss": 0.211,
+      "step": 2311
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.628920679322561e-07,
+      "loss": 0.2141,
+      "step": 2312
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.579410690818857e-07,
+      "loss": 0.2622,
+      "step": 2313
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.5300367807299e-07,
+      "loss": 0.2559,
+      "step": 2314
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.480799022548114e-07,
+      "loss": 0.2198,
+      "step": 2315
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.431697489563317e-07,
+      "loss": 0.2251,
+      "step": 2316
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.382732254862524e-07,
+      "loss": 0.2561,
+      "step": 2317
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.333903391329878e-07,
+      "loss": 0.2376,
+      "step": 2318
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.28521097164654e-07,
+      "loss": 0.2462,
+      "step": 2319
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.236655068290556e-07,
+      "loss": 0.2199,
+      "step": 2320
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.188235753536811e-07,
+      "loss": 0.21,
+      "step": 2321
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.139953099456843e-07,
+      "loss": 0.2578,
+      "step": 2322
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.091807177918776e-07,
+      "loss": 0.2714,
+      "step": 2323
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.043798060587271e-07,
+      "loss": 0.1998,
+      "step": 2324
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.995925818923222e-07,
+      "loss": 0.215,
+      "step": 2325
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.948190524183919e-07,
+      "loss": 0.219,
+      "step": 2326
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.900592247422734e-07,
+      "loss": 0.2241,
+      "step": 2327
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.853131059489139e-07,
+      "loss": 0.2086,
+      "step": 2328
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.805807031028512e-07,
+      "loss": 0.2356,
+      "step": 2329
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.758620232482083e-07,
+      "loss": 0.2148,
+      "step": 2330
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.711570734086804e-07,
+      "loss": 0.2178,
+      "step": 2331
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.664658605875275e-07,
+      "loss": 0.2641,
+      "step": 2332
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.617883917675639e-07,
+      "loss": 0.2306,
+      "step": 2333
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.571246739111426e-07,
+      "loss": 0.2465,
+      "step": 2334
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.524747139601474e-07,
+      "loss": 0.2753,
+      "step": 2335
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.478385188359916e-07,
+      "loss": 0.2385,
+      "step": 2336
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.432160954395895e-07,
+      "loss": 0.2047,
+      "step": 2337
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.38607450651364e-07,
+      "loss": 0.2192,
+      "step": 2338
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.340125913312235e-07,
+      "loss": 0.2525,
+      "step": 2339
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.294315243185579e-07,
+      "loss": 0.2366,
+      "step": 2340
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.248642564322339e-07,
+      "loss": 0.2332,
+      "step": 2341
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.203107944705679e-07,
+      "loss": 0.2431,
+      "step": 2342
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.15771145211337e-07,
+      "loss": 0.2189,
+      "step": 2343
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.112453154117482e-07,
+      "loss": 0.2506,
+      "step": 2344
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.067333118084429e-07,
+      "loss": 0.2299,
+      "step": 2345
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.022351411174866e-07,
+      "loss": 0.2206,
+      "step": 2346
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.977508100343466e-07,
+      "loss": 0.2165,
+      "step": 2347
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.932803252338971e-07,
+      "loss": 0.2233,
+      "step": 2348
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.888236933703985e-07,
+      "loss": 0.2117,
+      "step": 2349
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.843809210774921e-07,
+      "loss": 0.2389,
+      "step": 2350
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.799520149681893e-07,
+      "loss": 0.2205,
+      "step": 2351
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.7553698163486e-07,
+      "loss": 0.2532,
+      "step": 2352
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.711358276492296e-07,
+      "loss": 0.2259,
+      "step": 2353
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.667485595623602e-07,
+      "loss": 0.2231,
+      "step": 2354
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.623751839046455e-07,
+      "loss": 0.2313,
+      "step": 2355
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.580157071858029e-07,
+      "loss": 0.2184,
+      "step": 2356
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.536701358948561e-07,
+      "loss": 0.2318,
+      "step": 2357
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.493384765001376e-07,
+      "loss": 0.2271,
+      "step": 2358
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.450207354492654e-07,
+      "loss": 0.2431,
+      "step": 2359
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.407169191691465e-07,
+      "loss": 0.2098,
+      "step": 2360
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.364270340659584e-07,
+      "loss": 0.2317,
+      "step": 2361
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.32151086525139e-07,
+      "loss": 0.2816,
+      "step": 2362
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.278890829113859e-07,
+      "loss": 0.2202,
+      "step": 2363
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.236410295686357e-07,
+      "loss": 0.2507,
+      "step": 2364
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.194069328200669e-07,
+      "loss": 0.2006,
+      "step": 2365
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.15186798968077e-07,
+      "loss": 0.1991,
+      "step": 2366
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.109806342942837e-07,
+      "loss": 0.2354,
+      "step": 2367
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.067884450595151e-07,
+      "loss": 0.2444,
+      "step": 2368
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.026102375037857e-07,
+      "loss": 0.249,
+      "step": 2369
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.984460178463103e-07,
+      "loss": 0.198,
+      "step": 2370
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.942957922854775e-07,
+      "loss": 0.2197,
+      "step": 2371
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.901595669988458e-07,
+      "loss": 0.215,
+      "step": 2372
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.86037348143137e-07,
+      "loss": 0.2338,
+      "step": 2373
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.819291418542206e-07,
+      "loss": 0.2824,
+      "step": 2374
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.77834954247114e-07,
+      "loss": 0.2444,
+      "step": 2375
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.737547914159614e-07,
+      "loss": 0.2457,
+      "step": 2376
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.696886594340345e-07,
+      "loss": 0.2293,
+      "step": 2377
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.656365643537242e-07,
+      "loss": 0.2071,
+      "step": 2378
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.615985122065193e-07,
+      "loss": 0.2461,
+      "step": 2379
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.575745090030138e-07,
+      "loss": 0.2435,
+      "step": 2380
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.535645607328855e-07,
+      "loss": 0.2488,
+      "step": 2381
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.495686733648919e-07,
+      "loss": 0.2384,
+      "step": 2382
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.455868528468633e-07,
+      "loss": 0.2121,
+      "step": 2383
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.416191051056885e-07,
+      "loss": 0.2455,
+      "step": 2384
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.376654360473121e-07,
+      "loss": 0.2086,
+      "step": 2385
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.337258515567223e-07,
+      "loss": 0.2329,
+      "step": 2386
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.298003574979404e-07,
+      "loss": 0.2492,
+      "step": 2387
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.258889597140159e-07,
+      "loss": 0.2373,
+      "step": 2388
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.21991664027014e-07,
+      "loss": 0.2163,
+      "step": 2389
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.181084762380151e-07,
+      "loss": 0.254,
+      "step": 2390
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.142394021270924e-07,
+      "loss": 0.2363,
+      "step": 2391
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.103844474533182e-07,
+      "loss": 0.2284,
+      "step": 2392
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.065436179547434e-07,
+      "loss": 0.1981,
+      "step": 2393
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.02716919348395e-07,
+      "loss": 0.2085,
+      "step": 2394
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.989043573302655e-07,
+      "loss": 0.2294,
+      "step": 2395
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.951059375753064e-07,
+      "loss": 0.2526,
+      "step": 2396
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.913216657374198e-07,
+      "loss": 0.2503,
+      "step": 2397
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.875515474494474e-07,
+      "loss": 0.2126,
+      "step": 2398
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.837955883231615e-07,
+      "loss": 0.2728,
+      "step": 2399
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.800537939492645e-07,
+      "loss": 0.2413,
+      "step": 2400
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7632616989736534e-07,
+      "loss": 0.2135,
+      "step": 2401
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7261272171599146e-07,
+      "loss": 0.2067,
+      "step": 2402
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.6891345493256355e-07,
+      "loss": 0.2256,
+      "step": 2403
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.6522837505339037e-07,
+      "loss": 0.2532,
+      "step": 2404
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.61557487563673e-07,
+      "loss": 0.2401,
+      "step": 2405
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5790079792747966e-07,
+      "loss": 0.2498,
+      "step": 2406
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.542583115877475e-07,
+      "loss": 0.2402,
+      "step": 2407
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.506300339662717e-07,
+      "loss": 0.2699,
+      "step": 2408
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4701597046370007e-07,
+      "loss": 0.2617,
+      "step": 2409
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4341612645952047e-07,
+      "loss": 0.2076,
+      "step": 2410
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.398305073120546e-07,
+      "loss": 0.2181,
+      "step": 2411
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3625911835845523e-07,
+      "loss": 0.2295,
+      "step": 2412
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.327019649146891e-07,
+      "loss": 0.2319,
+      "step": 2413
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2915905227553623e-07,
+      "loss": 0.2558,
+      "step": 2414
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2563038571457605e-07,
+      "loss": 0.2389,
+      "step": 2415
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2211597048418374e-07,
+      "loss": 0.2393,
+      "step": 2416
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.186158118155281e-07,
+      "loss": 0.215,
+      "step": 2417
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1512991491854835e-07,
+      "loss": 0.2264,
+      "step": 2418
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1165828498195835e-07,
+      "loss": 0.251,
+      "step": 2419
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.08200927173239e-07,
+      "loss": 0.2539,
+      "step": 2420
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.047578466386204e-07,
+      "loss": 0.23,
+      "step": 2421
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.013290485030896e-07,
+      "loss": 0.2314,
+      "step": 2422
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.979145378703675e-07,
+      "loss": 0.2328,
+      "step": 2423
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.945143198229129e-07,
+      "loss": 0.2253,
+      "step": 2424
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9112839942190727e-07,
+      "loss": 0.2527,
+      "step": 2425
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8775678170724916e-07,
+      "loss": 0.1938,
+      "step": 2426
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.84399471697553e-07,
+      "loss": 0.203,
+      "step": 2427
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8105647439013016e-07,
+      "loss": 0.2424,
+      "step": 2428
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7772779476099364e-07,
+      "loss": 0.2196,
+      "step": 2429
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7441343776484116e-07,
+      "loss": 0.2707,
+      "step": 2430
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7111340833505094e-07,
+      "loss": 0.254,
+      "step": 2431
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.678277113836759e-07,
+      "loss": 0.2338,
+      "step": 2432
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.64556351801435e-07,
+      "loss": 0.2456,
+      "step": 2433
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.612993344577076e-07,
+      "loss": 0.2246,
+      "step": 2434
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5805666420052456e-07,
+      "loss": 0.2223,
+      "step": 2435
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5482834585655713e-07,
+      "loss": 0.2197,
+      "step": 2436
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.516143842311204e-07,
+      "loss": 0.2826,
+      "step": 2437
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.484147841081542e-07,
+      "loss": 0.2398,
+      "step": 2438
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4522955025022564e-07,
+      "loss": 0.2618,
+      "step": 2439
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.420586873985132e-07,
+      "loss": 0.2394,
+      "step": 2440
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.389022002728104e-07,
+      "loss": 0.2298,
+      "step": 2441
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3576009357150774e-07,
+      "loss": 0.2611,
+      "step": 2442
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3263237197158957e-07,
+      "loss": 0.2199,
+      "step": 2443
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.295190401286363e-07,
+      "loss": 0.2141,
+      "step": 2444
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.264201026767977e-07,
+      "loss": 0.2154,
+      "step": 2445
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.233355642288072e-07,
+      "loss": 0.2652,
+      "step": 2446
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.202654293759622e-07,
+      "loss": 0.2015,
+      "step": 2447
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.1720970268811823e-07,
+      "loss": 0.2295,
+      "step": 2448
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1416838871368925e-07,
+      "loss": 0.2366,
+      "step": 2449
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1114149197963185e-07,
+      "loss": 0.2522,
+      "step": 2450
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0812901699144526e-07,
+      "loss": 0.2099,
+      "step": 2451
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.051309682331605e-07,
+      "loss": 0.2362,
+      "step": 2452
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0214735016733444e-07,
+      "loss": 0.235,
+      "step": 2453
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9917816723504913e-07,
+      "loss": 0.2122,
+      "step": 2454
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9622342385589256e-07,
+      "loss": 0.2655,
+      "step": 2455
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.932831244279677e-07,
+      "loss": 0.2557,
+      "step": 2456
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9035727332787366e-07,
+      "loss": 0.2617,
+      "step": 2457
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8744587491069897e-07,
+      "loss": 0.2467,
+      "step": 2458
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.845489335100282e-07,
+      "loss": 0.2277,
+      "step": 2459
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8166645343792096e-07,
+      "loss": 0.2307,
+      "step": 2460
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.787984389849141e-07,
+      "loss": 0.1984,
+      "step": 2461
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.759448944200116e-07,
+      "loss": 0.2381,
+      "step": 2462
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7310582399067807e-07,
+      "loss": 0.2443,
+      "step": 2463
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.702812319228365e-07,
+      "loss": 0.2176,
+      "step": 2464
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.674711224208548e-07,
+      "loss": 0.2277,
+      "step": 2465
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6467549966754824e-07,
+      "loss": 0.2321,
+      "step": 2466
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.618943678241659e-07,
+      "loss": 0.2299,
+      "step": 2467
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5912773103038635e-07,
+      "loss": 0.1889,
+      "step": 2468
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.563755934043188e-07,
+      "loss": 0.2169,
+      "step": 2469
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.536379590424809e-07,
+      "loss": 0.244,
+      "step": 2470
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.509148320198107e-07,
+      "loss": 0.2176,
+      "step": 2471
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4820621638964813e-07,
+      "loss": 0.2369,
+      "step": 2472
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.455121161837337e-07,
+      "loss": 0.2161,
+      "step": 2473
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4283253541220186e-07,
+      "loss": 0.2476,
+      "step": 2474
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4016747806357657e-07,
+      "loss": 0.2137,
+      "step": 2475
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3751694810476232e-07,
+      "loss": 0.2237,
+      "step": 2476
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.348809494810389e-07,
+      "loss": 0.2291,
+      "step": 2477
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3225948611605985e-07,
+      "loss": 0.2038,
+      "step": 2478
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2965256191183949e-07,
+      "loss": 0.2268,
+      "step": 2479
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2706018074875046e-07,
+      "loss": 0.2157,
+      "step": 2480
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2448234648552391e-07,
+      "loss": 0.2445,
+      "step": 2481
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.219190629592316e-07,
+      "loss": 0.2142,
+      "step": 2482
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.193703339852904e-07,
+      "loss": 0.2105,
+      "step": 2483
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1683616335745117e-07,
+      "loss": 0.2031,
+      "step": 2484
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1431655484779435e-07,
+      "loss": 0.2426,
+      "step": 2485
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.118115122067288e-07,
+      "loss": 0.2186,
+      "step": 2486
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.093210391629763e-07,
+      "loss": 0.2052,
+      "step": 2487
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.068451394235793e-07,
+      "loss": 0.2165,
+      "step": 2488
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0438381667387875e-07,
+      "loss": 0.2722,
+      "step": 2489
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.019370745775273e-07,
+      "loss": 0.2362,
+      "step": 2490
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9950491677646843e-07,
+      "loss": 0.2576,
+      "step": 2491
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9708734689093846e-07,
+      "loss": 0.2175,
+      "step": 2492
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9468436851946104e-07,
+      "loss": 0.2093,
+      "step": 2493
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.922959852388395e-07,
+      "loss": 0.2305,
+      "step": 2494
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8992220060415346e-07,
+      "loss": 0.2089,
+      "step": 2495
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8756301814875088e-07,
+      "loss": 0.2108,
+      "step": 2496
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8521844138424617e-07,
+      "loss": 0.2416,
+      "step": 2497
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.828884738005121e-07,
+      "loss": 0.2371,
+      "step": 2498
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8057311886567785e-07,
+      "loss": 0.2241,
+      "step": 2499
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.782723800261199e-07,
+      "loss": 0.2334,
+      "step": 2500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7598626070645996e-07,
+      "loss": 0.2408,
+      "step": 2501
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7371476430955603e-07,
+      "loss": 0.2304,
+      "step": 2502
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.714578942165057e-07,
+      "loss": 0.2397,
+      "step": 2503
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6921565378662852e-07,
+      "loss": 0.2079,
+      "step": 2504
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.669880463574758e-07,
+      "loss": 0.2413,
+      "step": 2505
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6477507524481073e-07,
+      "loss": 0.2501,
+      "step": 2506
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6257674374261402e-07,
+      "loss": 0.231,
+      "step": 2507
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.603930551230759e-07,
+      "loss": 0.199,
+      "step": 2508
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.582240126365875e-07,
+      "loss": 0.224,
+      "step": 2509
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5606961951174394e-07,
+      "loss": 0.2485,
+      "step": 2510
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5392987895533118e-07,
+      "loss": 0.2229,
+      "step": 2511
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5180479415232707e-07,
+      "loss": 0.1978,
+      "step": 2512
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.496943682658958e-07,
+      "loss": 0.2161,
+      "step": 2513
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4759860443737784e-07,
+      "loss": 0.2321,
+      "step": 2514
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4551750578629232e-07,
+      "loss": 0.2416,
+      "step": 2515
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.43451075410328e-07,
+      "loss": 0.2554,
+      "step": 2516
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4139931638534222e-07,
+      "loss": 0.2121,
+      "step": 2517
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3936223176535202e-07,
+      "loss": 0.2239,
+      "step": 2518
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3733982458253302e-07,
+      "loss": 0.28,
+      "step": 2519
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3533209784721502e-07,
+      "loss": 0.2305,
+      "step": 2520
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3333905454787077e-07,
+      "loss": 0.2258,
+      "step": 2521
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.313606976511228e-07,
+      "loss": 0.2107,
+      "step": 2522
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.293970301017311e-07,
+      "loss": 0.2363,
+      "step": 2523
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2744805482258872e-07,
+      "loss": 0.2556,
+      "step": 2524
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2551377471472282e-07,
+      "loss": 0.2256,
+      "step": 2525
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.235941926572859e-07,
+      "loss": 0.215,
+      "step": 2526
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2168931150755015e-07,
+      "loss": 0.2065,
+      "step": 2527
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.197991341009086e-07,
+      "loss": 0.2185,
+      "step": 2528
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1792366325086624e-07,
+      "loss": 0.247,
+      "step": 2529
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.160629017490389e-07,
+      "loss": 0.2278,
+      "step": 2530
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1421685236514767e-07,
+      "loss": 0.2379,
+      "step": 2531
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1238551784701235e-07,
+      "loss": 0.1948,
+      "step": 2532
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.105689009205535e-07,
+      "loss": 0.2299,
+      "step": 2533
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0876700428978148e-07,
+      "loss": 0.2325,
+      "step": 2534
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.069798306367975e-07,
+      "loss": 0.225,
+      "step": 2535
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0520738262178809e-07,
+      "loss": 0.2122,
+      "step": 2536
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.034496628830195e-07,
+      "loss": 0.2238,
+      "step": 2537
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0170667403683665e-07,
+      "loss": 0.234,
+      "step": 2538
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.997841867765646e-08,
+      "loss": 0.2072,
+      "step": 2539
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.826489937796557e-08,
+      "loss": 0.2251,
+      "step": 2540
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.656611868831378e-08,
+      "loss": 0.2223,
+      "step": 2541
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.488207913731951e-08,
+      "loss": 0.2553,
+      "step": 2542
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.321278323165206e-08,
+      "loss": 0.2676,
+      "step": 2543
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.155823345603832e-08,
+      "loss": 0.2433,
+      "step": 2544
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.991843227325492e-08,
+      "loss": 0.2415,
+      "step": 2545
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.829338212412608e-08,
+      "loss": 0.2266,
+      "step": 2546
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.668308542751802e-08,
+      "loss": 0.2644,
+      "step": 2547
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.508754458033896e-08,
+      "loss": 0.2267,
+      "step": 2548
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.350676195752916e-08,
+      "loss": 0.2558,
+      "step": 2549
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.194073991206641e-08,
+      "loss": 0.2568,
+      "step": 2550
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.03894807749539e-08,
+      "loss": 0.253,
+      "step": 2551
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.885298685522235e-08,
+      "loss": 0.2347,
+      "step": 2552
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.733126043992233e-08,
+      "loss": 0.2673,
+      "step": 2553
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.582430379412641e-08,
+      "loss": 0.2159,
+      "step": 2554
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.433211916092143e-08,
+      "loss": 0.2055,
+      "step": 2555
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.285470876140288e-08,
+      "loss": 0.2487,
+      "step": 2556
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.139207479468169e-08,
+      "loss": 0.2169,
+      "step": 2557
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.994421943786744e-08,
+      "loss": 0.1996,
+      "step": 2558
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.851114484607513e-08,
+      "loss": 0.2436,
+      "step": 2559
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.709285315242064e-08,
+      "loss": 0.2304,
+      "step": 2560
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.568934646801083e-08,
+      "loss": 0.2126,
+      "step": 2561
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.430062688195015e-08,
+      "loss": 0.2144,
+      "step": 2562
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.292669646132953e-08,
+      "loss": 0.2312,
+      "step": 2563
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.156755725122754e-08,
+      "loss": 0.2338,
+      "step": 2564
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.022321127470698e-08,
+      "loss": 0.2392,
+      "step": 2565
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.8893660532807204e-08,
+      "loss": 0.2643,
+      "step": 2566
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.757890700455182e-08,
+      "loss": 0.2255,
+      "step": 2567
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.627895264693206e-08,
+      "loss": 0.2552,
+      "step": 2568
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.499379939491678e-08,
+      "loss": 0.2173,
+      "step": 2569
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.3723449161439124e-08,
+      "loss": 0.2175,
+      "step": 2570
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.246790383739875e-08,
+      "loss": 0.2113,
+      "step": 2571
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.1227165291660717e-08,
+      "loss": 0.2185,
+      "step": 2572
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.000123537104884e-08,
+      "loss": 0.2675,
+      "step": 2573
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8790115900343436e-08,
+      "loss": 0.2118,
+      "step": 2574
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.759380868228247e-08,
+      "loss": 0.2254,
+      "step": 2575
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.641231549755376e-08,
+      "loss": 0.1884,
+      "step": 2576
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5245638104794986e-08,
+      "loss": 0.2199,
+      "step": 2577
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.409377824059147e-08,
+      "loss": 0.2575,
+      "step": 2578
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.295673761947172e-08,
+      "loss": 0.2236,
+      "step": 2579
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.183451793390747e-08,
+      "loss": 0.242,
+      "step": 2580
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.072712085430808e-08,
+      "loss": 0.2309,
+      "step": 2581
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.963454802902056e-08,
+      "loss": 0.2319,
+      "step": 2582
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8556801084326244e-08,
+      "loss": 0.2356,
+      "step": 2583
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7493881624437456e-08,
+      "loss": 0.2209,
+      "step": 2584
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6445791231497496e-08,
+      "loss": 0.2411,
+      "step": 2585
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.541253146557511e-08,
+      "loss": 0.2166,
+      "step": 2586
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.4394103864666684e-08,
+      "loss": 0.2111,
+      "step": 2587
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.339050994468629e-08,
+      "loss": 0.2417,
+      "step": 2588
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.2401751199472306e-08,
+      "loss": 0.2263,
+      "step": 2589
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.1427829100779686e-08,
+      "loss": 0.2091,
+      "step": 2590
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0468745098281016e-08,
+      "loss": 0.2039,
+      "step": 2591
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9524500619558806e-08,
+      "loss": 0.2304,
+      "step": 2592
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8595097070108768e-08,
+      "loss": 0.223,
+      "step": 2593
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.768053583333874e-08,
+      "loss": 0.2438,
+      "step": 2594
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6780818270562002e-08,
+      "loss": 0.2114,
+      "step": 2595
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5895945720993963e-08,
+      "loss": 0.2216,
+      "step": 2596
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.502591950175992e-08,
+      "loss": 0.243,
+      "step": 2597
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.417074090788063e-08,
+      "loss": 0.2112,
+      "step": 2598
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.333041121228008e-08,
+      "loss": 0.2191,
+      "step": 2599
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.250493166577772e-08,
+      "loss": 0.2232,
+      "step": 2600
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1694303497090675e-08,
+      "loss": 0.2341,
+      "step": 2601
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0898527912828202e-08,
+      "loss": 0.266,
+      "step": 2602
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0117606097492803e-08,
+      "loss": 0.1988,
+      "step": 2603
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.935153921347688e-08,
+      "loss": 0.2276,
+      "step": 2604
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.860032840106163e-08,
+      "loss": 0.2071,
+      "step": 2605
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.786397477841484e-08,
+      "loss": 0.2143,
+      "step": 2606
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7142479441589754e-08,
+      "loss": 0.2289,
+      "step": 2607
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6435843464522873e-08,
+      "loss": 0.2142,
+      "step": 2608
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5744067899032823e-08,
+      "loss": 0.2433,
+      "step": 2609
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5067153774820375e-08,
+      "loss": 0.2151,
+      "step": 2610
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4405102099462887e-08,
+      "loss": 0.2029,
+      "step": 2611
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3757913858416516e-08,
+      "loss": 0.2683,
+      "step": 2612
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.312559001501179e-08,
+      "loss": 0.2903,
+      "step": 2613
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2508131510456934e-08,
+      "loss": 0.2275,
+      "step": 2614
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.190553926382898e-08,
+      "loss": 0.2435,
+      "step": 2615
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1317814172080444e-08,
+      "loss": 0.2077,
+      "step": 2616
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.074495711003265e-08,
+      "loss": 0.2877,
+      "step": 2617
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.018696893037685e-08,
+      "loss": 0.2311,
+      "step": 2618
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.64385046366978e-09,
+      "loss": 0.2236,
+      "step": 2619
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.115602518338096e-09,
+      "loss": 0.218,
+      "step": 2620
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.602225880671721e-09,
+      "loss": 0.2493,
+      "step": 2621
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.103721314826063e-09,
+      "loss": 0.2301,
+      "step": 2622
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.620089562817568e-09,
+      "loss": 0.2112,
+      "step": 2623
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.1513313445270575e-09,
+      "loss": 0.24,
+      "step": 2624
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.697447357695286e-09,
+      "loss": 0.2279,
+      "step": 2625
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.25843827792405e-09,
+      "loss": 0.2174,
+      "step": 2626
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.834304758670639e-09,
+      "loss": 0.1921,
+      "step": 2627
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.425047431254493e-09,
+      "loss": 0.2235,
+      "step": 2628
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.030666904848325e-09,
+      "loss": 0.2229,
+      "step": 2629
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.651163766484779e-09,
+      "loss": 0.2211,
+      "step": 2630
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.28653858104755e-09,
+      "loss": 0.2461,
+      "step": 2631
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.936791891278047e-09,
+      "loss": 0.2045,
+      "step": 2632
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.6019242177698368e-09,
+      "loss": 0.2264,
+      "step": 2633
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.2819360589675386e-09,
+      "loss": 0.2229,
+      "step": 2634
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9768278911723737e-09,
+      "loss": 0.2479,
+      "step": 2635
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.6866001685310616e-09,
+      "loss": 0.2002,
+      "step": 2636
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.411253323046925e-09,
+      "loss": 0.2429,
+      "step": 2637
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1507877645687846e-09,
+      "loss": 0.1941,
+      "step": 2638
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.905203880795403e-09,
+      "loss": 0.2094,
+      "step": 2639
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6745020372777033e-09,
+      "loss": 0.2108,
+      "step": 2640
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.458682577410997e-09,
+      "loss": 0.2298,
+      "step": 2641
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.2577458224394267e-09,
+      "loss": 0.2179,
+      "step": 2642
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0716920714570755e-09,
+      "loss": 0.279,
+      "step": 2643
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.00521601400195e-10,
+      "loss": 0.2297,
+      "step": 2644
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.442346670549771e-10,
+      "loss": 0.1712,
+      "step": 2645
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.028315010531139e-10,
+      "loss": 0.2582,
+      "step": 2646
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.763123138695758e-10,
+      "loss": 0.2595,
+      "step": 2647
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.646772938292742e-10,
+      "loss": 0.2005,
+      "step": 2648
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.6792660709706875e-10,
+      "loss": 0.233,
+      "step": 2649
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8606039768775952e-10,
+      "loss": 0.2189,
+      "step": 2650
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.190787874549848e-10,
+      "loss": 0.2492,
+      "step": 2651
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.698187610343354e-11,
+      "loss": 0.2347,
+      "step": 2652
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.976974117552267e-11,
+      "loss": 0.2094,
+      "step": 2653
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.442438063609558e-12,
+      "loss": 0.2053,
+      "step": 2654
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.2228,
+      "step": 2655
+    },
+    {
+      "epoch": 3.0,
+      "step": 2655,
+      "total_flos": 1.202832417476313e+18,
+      "train_loss": 0.5172335097064631,
+      "train_runtime": 25247.7542,
+      "train_samples_per_second": 6.733,
+      "train_steps_per_second": 0.105
+    }
+  ],
+  "max_steps": 2655,
+  "num_train_epochs": 3,
+  "total_flos": 1.202832417476313e+18,
+  "trial_name": null,
+  "trial_params": null
+}