diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,19761 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.996349254639489,
+  "global_step": 3284,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.787878787878788e-07,
+      "loss": 1.4658,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.575757575757576e-07,
+      "loss": 1.4105,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1363636363636364e-06,
+      "loss": 1.5376,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5151515151515152e-06,
+      "loss": 1.4621,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8939393939393941e-06,
+      "loss": 1.4424,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2727272727272728e-06,
+      "loss": 1.3302,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.651515151515152e-06,
+      "loss": 1.3399,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0303030303030305e-06,
+      "loss": 1.2243,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.409090909090909e-06,
+      "loss": 1.1229,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7878787878787882e-06,
+      "loss": 0.9401,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 0.8221,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5454545454545455e-06,
+      "loss": 0.7782,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.924242424242424e-06,
+      "loss": 0.7178,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.303030303030304e-06,
+      "loss": 0.6973,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.681818181818182e-06,
+      "loss": 0.6625,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.060606060606061e-06,
+      "loss": 0.627,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.43939393939394e-06,
+      "loss": 0.582,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.818181818181818e-06,
+      "loss": 0.5465,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.196969696969698e-06,
+      "loss": 0.5315,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.5757575757575764e-06,
+      "loss": 0.5779,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.954545454545455e-06,
+      "loss": 0.4977,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.4814,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.712121212121212e-06,
+      "loss": 0.4637,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 0.4718,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.46969696969697e-06,
+      "loss": 0.4661,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.848484848484848e-06,
+      "loss": 0.4551,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0227272727272729e-05,
+      "loss": 0.4872,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0606060606060607e-05,
+      "loss": 0.4514,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0984848484848486e-05,
+      "loss": 0.4532,
+      "step": 29
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1363636363636365e-05,
+      "loss": 0.4611,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1742424242424243e-05,
+      "loss": 0.4541,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2121212121212122e-05,
+      "loss": 0.4964,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.25e-05,
+      "loss": 0.4659,
+      "step": 33
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.287878787878788e-05,
+      "loss": 0.4854,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3257575757575758e-05,
+      "loss": 0.4623,
+      "step": 35
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3636363636363637e-05,
+      "loss": 0.4888,
+      "step": 36
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4015151515151515e-05,
+      "loss": 0.4599,
+      "step": 37
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4393939393939396e-05,
+      "loss": 0.4943,
+      "step": 38
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4772727272727274e-05,
+      "loss": 0.4305,
+      "step": 39
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5151515151515153e-05,
+      "loss": 0.4147,
+      "step": 40
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.553030303030303e-05,
+      "loss": 0.4876,
+      "step": 41
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.590909090909091e-05,
+      "loss": 0.4417,
+      "step": 42
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.628787878787879e-05,
+      "loss": 0.3859,
+      "step": 43
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.4474,
+      "step": 44
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7045454545454546e-05,
+      "loss": 0.4073,
+      "step": 45
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7424242424242425e-05,
+      "loss": 0.4758,
+      "step": 46
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7803030303030303e-05,
+      "loss": 0.414,
+      "step": 47
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 0.4243,
+      "step": 48
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.856060606060606e-05,
+      "loss": 0.4426,
+      "step": 49
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.893939393939394e-05,
+      "loss": 0.4245,
+      "step": 50
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9318181818181818e-05,
+      "loss": 0.4364,
+      "step": 51
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9696969696969697e-05,
+      "loss": 0.4285,
+      "step": 52
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0075757575757575e-05,
+      "loss": 0.3968,
+      "step": 53
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.0454545454545457e-05,
+      "loss": 0.4738,
+      "step": 54
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.0833333333333336e-05,
+      "loss": 0.4445,
+      "step": 55
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1212121212121215e-05,
+      "loss": 0.4266,
+      "step": 56
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1590909090909093e-05,
+      "loss": 0.4481,
+      "step": 57
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1969696969696972e-05,
+      "loss": 0.4377,
+      "step": 58
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.234848484848485e-05,
+      "loss": 0.4621,
+      "step": 59
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.272727272727273e-05,
+      "loss": 0.4731,
+      "step": 60
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3106060606060605e-05,
+      "loss": 0.4484,
+      "step": 61
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3484848484848487e-05,
+      "loss": 0.4442,
+      "step": 62
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.3863636363636365e-05,
+      "loss": 0.4277,
+      "step": 63
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4242424242424244e-05,
+      "loss": 0.4432,
+      "step": 64
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.4621212121212123e-05,
+      "loss": 0.4079,
+      "step": 65
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5e-05,
+      "loss": 0.4114,
+      "step": 66
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.537878787878788e-05,
+      "loss": 0.4372,
+      "step": 67
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.575757575757576e-05,
+      "loss": 0.4175,
+      "step": 68
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6136363636363637e-05,
+      "loss": 0.4316,
+      "step": 69
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.6515151515151516e-05,
+      "loss": 0.4543,
+      "step": 70
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.6893939393939394e-05,
+      "loss": 0.4581,
+      "step": 71
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7272727272727273e-05,
+      "loss": 0.428,
+      "step": 72
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7651515151515152e-05,
+      "loss": 0.4652,
+      "step": 73
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.803030303030303e-05,
+      "loss": 0.4003,
+      "step": 74
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8409090909090912e-05,
+      "loss": 0.4271,
+      "step": 75
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.878787878787879e-05,
+      "loss": 0.4497,
+      "step": 76
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.916666666666667e-05,
+      "loss": 0.4314,
+      "step": 77
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.954545454545455e-05,
+      "loss": 0.405,
+      "step": 78
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9924242424242427e-05,
+      "loss": 0.4641,
+      "step": 79
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0303030303030306e-05,
+      "loss": 0.4236,
+      "step": 80
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.068181818181818e-05,
+      "loss": 0.4242,
+      "step": 81
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.106060606060606e-05,
+      "loss": 0.4344,
+      "step": 82
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.143939393939394e-05,
+      "loss": 0.4494,
+      "step": 83
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.181818181818182e-05,
+      "loss": 0.4401,
+      "step": 84
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2196969696969696e-05,
+      "loss": 0.4209,
+      "step": 85
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.257575757575758e-05,
+      "loss": 0.4217,
+      "step": 86
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.295454545454545e-05,
+      "loss": 0.4484,
+      "step": 87
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.4175,
+      "step": 88
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.371212121212121e-05,
+      "loss": 0.4998,
+      "step": 89
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.409090909090909e-05,
+      "loss": 0.4104,
+      "step": 90
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.4469696969696974e-05,
+      "loss": 0.415,
+      "step": 91
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.484848484848485e-05,
+      "loss": 0.4673,
+      "step": 92
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.522727272727273e-05,
+      "loss": 0.4641,
+      "step": 93
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.560606060606061e-05,
+      "loss": 0.4412,
+      "step": 94
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.598484848484849e-05,
+      "loss": 0.4367,
+      "step": 95
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.6363636363636364e-05,
+      "loss": 0.4003,
+      "step": 96
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.6742424242424246e-05,
+      "loss": 0.407,
+      "step": 97
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.712121212121212e-05,
+      "loss": 0.43,
+      "step": 98
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.4702,
+      "step": 99
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.787878787878788e-05,
+      "loss": 0.4386,
+      "step": 100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.825757575757576e-05,
+      "loss": 0.4287,
+      "step": 101
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.8636363636363636e-05,
+      "loss": 0.4738,
+      "step": 102
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.901515151515152e-05,
+      "loss": 0.4126,
+      "step": 103
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.939393939393939e-05,
+      "loss": 0.4542,
+      "step": 104
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9772727272727275e-05,
+      "loss": 0.4865,
+      "step": 105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.015151515151515e-05,
+      "loss": 0.4388,
+      "step": 106
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.053030303030303e-05,
+      "loss": 0.4383,
+      "step": 107
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.0909090909090915e-05,
+      "loss": 0.4309,
+      "step": 108
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.128787878787879e-05,
+      "loss": 0.4881,
+      "step": 109
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.438,
+      "step": 110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.204545454545455e-05,
+      "loss": 0.3896,
+      "step": 111
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.242424242424243e-05,
+      "loss": 0.4361,
+      "step": 112
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2803030303030305e-05,
+      "loss": 0.4639,
+      "step": 113
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.318181818181819e-05,
+      "loss": 0.4402,
+      "step": 114
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.356060606060606e-05,
+      "loss": 0.4074,
+      "step": 115
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3939393939393944e-05,
+      "loss": 0.461,
+      "step": 116
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.431818181818182e-05,
+      "loss": 0.4746,
+      "step": 117
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.46969696969697e-05,
+      "loss": 0.4736,
+      "step": 118
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.5075757575757577e-05,
+      "loss": 0.4344,
+      "step": 119
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.545454545454546e-05,
+      "loss": 0.4151,
+      "step": 120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.5833333333333334e-05,
+      "loss": 0.5076,
+      "step": 121
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.621212121212121e-05,
+      "loss": 0.4392,
+      "step": 122
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.659090909090909e-05,
+      "loss": 0.4609,
+      "step": 123
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.696969696969697e-05,
+      "loss": 0.4597,
+      "step": 124
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7348484848484855e-05,
+      "loss": 0.4404,
+      "step": 125
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.772727272727273e-05,
+      "loss": 0.4331,
+      "step": 126
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.810606060606061e-05,
+      "loss": 0.4589,
+      "step": 127
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.848484848484849e-05,
+      "loss": 0.4648,
+      "step": 128
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.886363636363637e-05,
+      "loss": 0.4559,
+      "step": 129
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9242424242424245e-05,
+      "loss": 0.4689,
+      "step": 130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.962121212121213e-05,
+      "loss": 0.4991,
+      "step": 131
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5e-05,
+      "loss": 0.4031,
+      "step": 132
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999987582410334e-05,
+      "loss": 0.4193,
+      "step": 133
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999950329653655e-05,
+      "loss": 0.4864,
+      "step": 134
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999988824176699e-05,
+      "loss": 0.4255,
+      "step": 135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9999801318812e-05,
+      "loss": 0.4295,
+      "step": 136
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999968956087504e-05,
+      "loss": 0.4498,
+      "step": 137
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999955296806713e-05,
+      "loss": 0.4425,
+      "step": 138
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999939154052397e-05,
+      "loss": 0.4334,
+      "step": 139
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999920527840592e-05,
+      "loss": 0.4848,
+      "step": 140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9998994181898005e-05,
+      "loss": 0.4334,
+      "step": 141
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999875825120994e-05,
+      "loss": 0.4362,
+      "step": 142
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.99984974865761e-05,
+      "loss": 0.5044,
+      "step": 143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999821188825553e-05,
+      "loss": 0.4477,
+      "step": 144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9997901456531935e-05,
+      "loss": 0.4627,
+      "step": 145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9997566191713715e-05,
+      "loss": 0.4913,
+      "step": 146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9997206094133915e-05,
+      "loss": 0.4563,
+      "step": 147
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999682116415026e-05,
+      "loss": 0.4518,
+      "step": 148
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9996411402145146e-05,
+      "loss": 0.4371,
+      "step": 149
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9995976808525625e-05,
+      "loss": 0.427,
+      "step": 150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9995517383723434e-05,
+      "loss": 0.4498,
+      "step": 151
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999503312819497e-05,
+      "loss": 0.4325,
+      "step": 152
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999452404242129e-05,
+      "loss": 0.4564,
+      "step": 153
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999399012690813e-05,
+      "loss": 0.4516,
+      "step": 154
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999343138218587e-05,
+      "loss": 0.4222,
+      "step": 155
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.99928478088096e-05,
+      "loss": 0.4925,
+      "step": 156
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999223940735901e-05,
+      "loss": 0.4041,
+      "step": 157
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999160617843851e-05,
+      "loss": 0.4166,
+      "step": 158
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9990948122677165e-05,
+      "loss": 0.4351,
+      "step": 159
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9990265240728674e-05,
+      "loss": 0.4401,
+      "step": 160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9989557533271414e-05,
+      "loss": 0.4433,
+      "step": 161
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9988825001008443e-05,
+      "loss": 0.421,
+      "step": 162
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998806764466746e-05,
+      "loss": 0.4256,
+      "step": 163
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998728546500082e-05,
+      "loss": 0.3848,
+      "step": 164
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9986478462785544e-05,
+      "loss": 0.5373,
+      "step": 165
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9985646638823335e-05,
+      "loss": 0.4658,
+      "step": 166
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9984789993940505e-05,
+      "loss": 0.4071,
+      "step": 167
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998390852898807e-05,
+      "loss": 0.4111,
+      "step": 168
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998300224484168e-05,
+      "loss": 0.4877,
+      "step": 169
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998207114240164e-05,
+      "loss": 0.4763,
+      "step": 170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998111522259291e-05,
+      "loss": 0.479,
+      "step": 171
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998013448636512e-05,
+      "loss": 0.4539,
+      "step": 172
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997912893469253e-05,
+      "loss": 0.4458,
+      "step": 173
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9978098568574074e-05,
+      "loss": 0.4223,
+      "step": 174
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9977043389033305e-05,
+      "loss": 0.4448,
+      "step": 175
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9975963397118464e-05,
+      "loss": 0.4353,
+      "step": 176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9974858593902416e-05,
+      "loss": 0.4446,
+      "step": 177
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997372898048268e-05,
+      "loss": 0.4281,
+      "step": 178
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9972574557981435e-05,
+      "loss": 0.4568,
+      "step": 179
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9971395327545466e-05,
+      "loss": 0.4378,
+      "step": 180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9970191290346245e-05,
+      "loss": 0.4759,
+      "step": 181
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.996896244757988e-05,
+      "loss": 0.4203,
+      "step": 182
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9967708800467095e-05,
+      "loss": 0.4187,
+      "step": 183
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.996643035025328e-05,
+      "loss": 0.4376,
+      "step": 184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996512709820845e-05,
+      "loss": 0.4162,
+      "step": 185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996379904562728e-05,
+      "loss": 0.4021,
+      "step": 186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996244619382906e-05,
+      "loss": 0.4718,
+      "step": 187
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9961068544157714e-05,
+      "loss": 0.4693,
+      "step": 188
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.995966609798182e-05,
+      "loss": 0.4414,
+      "step": 189
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9958238856694574e-05,
+      "loss": 0.4461,
+      "step": 190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9956786821713795e-05,
+      "loss": 0.4587,
+      "step": 191
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9955309994481967e-05,
+      "loss": 0.4159,
+      "step": 192
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.995380837646617e-05,
+      "loss": 0.4473,
+      "step": 193
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995228196915813e-05,
+      "loss": 0.4405,
+      "step": 194
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9950730774074176e-05,
+      "loss": 0.4873,
+      "step": 195
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9949154792755286e-05,
+      "loss": 0.4017,
+      "step": 196
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.994755402676705e-05,
+      "loss": 0.4477,
+      "step": 197
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.994592847769968e-05,
+      "loss": 0.4709,
+      "step": 198
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.994427814716801e-05,
+      "loss": 0.4472,
+      "step": 199
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.994260303681148e-05,
+      "loss": 0.432,
+      "step": 200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.994090314829418e-05,
+      "loss": 0.4405,
+      "step": 201
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.993917848330477e-05,
+      "loss": 0.4942,
+      "step": 202
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.993742904355655e-05,
+      "loss": 0.3964,
+      "step": 203
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.993565483078743e-05,
+      "loss": 0.4617,
+      "step": 204
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9933855846759915e-05,
+      "loss": 0.4231,
+      "step": 205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9932032093261136e-05,
+      "loss": 0.3823,
+      "step": 206
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.993018357210283e-05,
+      "loss": 0.4488,
+      "step": 207
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.992831028512133e-05,
+      "loss": 0.4698,
+      "step": 208
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9926412234177555e-05,
+      "loss": 0.4307,
+      "step": 209
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9924489421157056e-05,
+      "loss": 0.4313,
+      "step": 210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.992254184796997e-05,
+      "loss": 0.4422,
+      "step": 211
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.992056951655103e-05,
+      "loss": 0.4072,
+      "step": 212
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9918572428859554e-05,
+      "loss": 0.4573,
+      "step": 213
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.991655058687948e-05,
+      "loss": 0.4392,
+      "step": 214
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.991450399261931e-05,
+      "loss": 0.4244,
+      "step": 215
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.991243264811215e-05,
+      "loss": 0.452,
+      "step": 216
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.991033655541568e-05,
+      "loss": 0.4025,
+      "step": 217
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9908215716612186e-05,
+      "loss": 0.4053,
+      "step": 218
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9906070133808516e-05,
+      "loss": 0.4285,
+      "step": 219
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.99038998091361e-05,
+      "loss": 0.4585,
+      "step": 220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.990170474475096e-05,
+      "loss": 0.4529,
+      "step": 221
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.989948494283371e-05,
+      "loss": 0.4486,
+      "step": 222
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9897240405589485e-05,
+      "loss": 0.4912,
+      "step": 223
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9894971135248044e-05,
+      "loss": 0.3463,
+      "step": 224
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9892677134063694e-05,
+      "loss": 0.4249,
+      "step": 225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.98903584043153e-05,
+      "loss": 0.4179,
+      "step": 226
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.988801494830632e-05,
+      "loss": 0.4628,
+      "step": 227
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.988564676836475e-05,
+      "loss": 0.4645,
+      "step": 228
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9883253866843166e-05,
+      "loss": 0.4621,
+      "step": 229
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.988083624611868e-05,
+      "loss": 0.4505,
+      "step": 230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.987839390859299e-05,
+      "loss": 0.4373,
+      "step": 231
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.987592685669232e-05,
+      "loss": 0.4974,
+      "step": 232
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.987343509286746e-05,
+      "loss": 0.4655,
+      "step": 233
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.987091861959375e-05,
+      "loss": 0.4513,
+      "step": 234
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.986837743937107e-05,
+      "loss": 0.452,
+      "step": 235
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.986581155472384e-05,
+      "loss": 0.4372,
+      "step": 236
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.986322096820104e-05,
+      "loss": 0.3996,
+      "step": 237
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.986060568237617e-05,
+      "loss": 0.4216,
+      "step": 238
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.985796569984727e-05,
+      "loss": 0.4128,
+      "step": 239
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9855301023236925e-05,
+      "loss": 0.4206,
+      "step": 240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9852611655192244e-05,
+      "loss": 0.4566,
+      "step": 241
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9849897598384864e-05,
+      "loss": 0.3869,
+      "step": 242
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9847158855510945e-05,
+      "loss": 0.4776,
+      "step": 243
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.984439542929117e-05,
+      "loss": 0.4835,
+      "step": 244
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9841607322470755e-05,
+      "loss": 0.4379,
+      "step": 245
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.983879453781942e-05,
+      "loss": 0.4292,
+      "step": 246
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9835957078131414e-05,
+      "loss": 0.4281,
+      "step": 247
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.983309494622548e-05,
+      "loss": 0.4141,
+      "step": 248
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9830208144944886e-05,
+      "loss": 0.4367,
+      "step": 249
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.98272966771574e-05,
+      "loss": 0.4056,
+      "step": 250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.982436054575529e-05,
+      "loss": 0.4086,
+      "step": 251
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.982139975365533e-05,
+      "loss": 0.4598,
+      "step": 252
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.98184143037988e-05,
+      "loss": 0.4494,
+      "step": 253
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.981540419915147e-05,
+      "loss": 0.4107,
+      "step": 254
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9812369442703585e-05,
+      "loss": 0.4387,
+      "step": 255
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9809310037469903e-05,
+      "loss": 0.401,
+      "step": 256
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.980622598648965e-05,
+      "loss": 0.4675,
+      "step": 257
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9803117292826565e-05,
+      "loss": 0.4159,
+      "step": 258
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.979998395956883e-05,
+      "loss": 0.387,
+      "step": 259
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.979682598982912e-05,
+      "loss": 0.4628,
+      "step": 260
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.979364338674459e-05,
+      "loss": 0.4509,
+      "step": 261
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.979043615347686e-05,
+      "loss": 0.4314,
+      "step": 262
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.978720429321203e-05,
+      "loss": 0.4346,
+      "step": 263
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.978394780916063e-05,
+      "loss": 0.4339,
+      "step": 264
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9780666704557685e-05,
+      "loss": 0.4148,
+      "step": 265
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.977736098266268e-05,
+      "loss": 0.4594,
+      "step": 266
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9774030646759525e-05,
+      "loss": 0.4244,
+      "step": 267
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.977067570015661e-05,
+      "loss": 0.4603,
+      "step": 268
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.976729614618675e-05,
+      "loss": 0.4358,
+      "step": 269
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.976389198820723e-05,
+      "loss": 0.4209,
+      "step": 270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.976046322959978e-05,
+      "loss": 0.4281,
+      "step": 271
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.975700987377051e-05,
+      "loss": 0.3622,
+      "step": 272
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9753531924150054e-05,
+      "loss": 0.4267,
+      "step": 273
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9750029384193405e-05,
+      "loss": 0.4152,
+      "step": 274
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9746502257380015e-05,
+      "loss": 0.4366,
+      "step": 275
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.974295054721377e-05,
+      "loss": 0.4357,
+      "step": 276
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.973937425722295e-05,
+      "loss": 0.4563,
+      "step": 277
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.973577339096027e-05,
+      "loss": 0.4576,
+      "step": 278
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.973214795200287e-05,
+      "loss": 0.4361,
+      "step": 279
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.972849794395227e-05,
+      "loss": 0.4453,
+      "step": 280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9724823370434415e-05,
+      "loss": 0.4249,
+      "step": 281
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.972112423509966e-05,
+      "loss": 0.485,
+      "step": 282
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9717400541622756e-05,
+      "loss": 0.4678,
+      "step": 283
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.971365229370284e-05,
+      "loss": 0.4624,
+      "step": 284
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.970987949506344e-05,
+      "loss": 0.4445,
+      "step": 285
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9706082149452505e-05,
+      "loss": 0.3774,
+      "step": 286
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.970226026064232e-05,
+      "loss": 0.4512,
+      "step": 287
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.969841383242959e-05,
+      "loss": 0.439,
+      "step": 288
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.969454286863538e-05,
+      "loss": 0.4002,
+      "step": 289
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.969064737310515e-05,
+      "loss": 0.4184,
+      "step": 290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9686727349708686e-05,
+      "loss": 0.396,
+      "step": 291
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9682782802340184e-05,
+      "loss": 0.4441,
+      "step": 292
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.967881373491818e-05,
+      "loss": 0.4065,
+      "step": 293
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.967482015138558e-05,
+      "loss": 0.4375,
+      "step": 294
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.967080205570963e-05,
+      "loss": 0.4531,
+      "step": 295
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.966675945188194e-05,
+      "loss": 0.4626,
+      "step": 296
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.966269234391846e-05,
+      "loss": 0.4667,
+      "step": 297
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9658600735859486e-05,
+      "loss": 0.4319,
+      "step": 298
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.965448463176965e-05,
+      "loss": 0.4545,
+      "step": 299
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.965034403573792e-05,
+      "loss": 0.4688,
+      "step": 300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.96461789518776e-05,
+      "loss": 0.4018,
+      "step": 301
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.96419893843263e-05,
+      "loss": 0.3939,
+      "step": 302
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.963777533724597e-05,
+      "loss": 0.3895,
+      "step": 303
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.963353681482289e-05,
+      "loss": 0.4278,
+      "step": 304
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.962927382126762e-05,
+      "loss": 0.4116,
+      "step": 305
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.962498636081505e-05,
+      "loss": 0.4413,
+      "step": 306
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.962067443772439e-05,
+      "loss": 0.4251,
+      "step": 307
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9616338056279124e-05,
+      "loss": 0.3939,
+      "step": 308
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9611977220787035e-05,
+      "loss": 0.4047,
+      "step": 309
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9607591935580224e-05,
+      "loss": 0.4373,
+      "step": 310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9603182205015066e-05,
+      "loss": 0.4221,
+      "step": 311
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9598748033472196e-05,
+      "loss": 0.41,
+      "step": 312
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9594289425356585e-05,
+      "loss": 0.4205,
+      "step": 313
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.958980638509742e-05,
+      "loss": 0.4122,
+      "step": 314
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9585298917148196e-05,
+      "loss": 0.411,
+      "step": 315
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.958076702598666e-05,
+      "loss": 0.4455,
+      "step": 316
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.957621071611483e-05,
+      "loss": 0.4439,
+      "step": 317
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9571629992058985e-05,
+      "loss": 0.4083,
+      "step": 318
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.956702485836962e-05,
+      "loss": 0.4348,
+      "step": 319
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9562395319621545e-05,
+      "loss": 0.4797,
+      "step": 320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.955774138041375e-05,
+      "loss": 0.4264,
+      "step": 321
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.95530630453695e-05,
+      "loss": 0.4087,
+      "step": 322
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.954836031913629e-05,
+      "loss": 0.4378,
+      "step": 323
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9543633206385834e-05,
+      "loss": 0.4439,
+      "step": 324
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.953888171181409e-05,
+      "loss": 0.4132,
+      "step": 325
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.953410584014122e-05,
+      "loss": 0.4077,
+      "step": 326
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.952930559611161e-05,
+      "loss": 0.4332,
+      "step": 327
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.952448098449386e-05,
+      "loss": 0.4609,
+      "step": 328
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.951963201008076e-05,
+      "loss": 0.4743,
+      "step": 329
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.951475867768934e-05,
+      "loss": 0.4782,
+      "step": 330
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.950986099216078e-05,
+      "loss": 0.4098,
+      "step": 331
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.95049389583605e-05,
+      "loss": 0.4137,
+      "step": 332
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.949999258117806e-05,
+      "loss": 0.4455,
+      "step": 333
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.949502186552724e-05,
+      "loss": 0.437,
+      "step": 334
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.949002681634598e-05,
+      "loss": 0.4045,
+      "step": 335
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9485007438596395e-05,
+      "loss": 0.4006,
+      "step": 336
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.947996373726478e-05,
+      "loss": 0.4877,
+      "step": 337
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.947489571736158e-05,
+      "loss": 0.3569,
+      "step": 338
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.94698033839214e-05,
+      "loss": 0.4306,
+      "step": 339
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9464686742003006e-05,
+      "loss": 0.4108,
+      "step": 340
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.94595457966893e-05,
+      "loss": 0.3941,
+      "step": 341
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.945438055308733e-05,
+      "loss": 0.3939,
+      "step": 342
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.94491910163283e-05,
+      "loss": 0.4389,
+      "step": 343
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9443977191567526e-05,
+      "loss": 0.3884,
+      "step": 344
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.943873908398445e-05,
+      "loss": 0.3798,
+      "step": 345
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.943347669878267e-05,
+      "loss": 0.4011,
+      "step": 346
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9428190041189845e-05,
+      "loss": 0.4599,
+      "step": 347
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.94228791164578e-05,
+      "loss": 0.4462,
+      "step": 348
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.941754392986244e-05,
+      "loss": 0.4376,
+      "step": 349
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.941218448670378e-05,
+      "loss": 0.4147,
+      "step": 350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.940680079230593e-05,
+      "loss": 0.4124,
+      "step": 351
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9401392852017095e-05,
+      "loss": 0.3963,
+      "step": 352
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.939596067120954e-05,
+      "loss": 0.4673,
+      "step": 353
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9390504255279655e-05,
+      "loss": 0.4381,
+      "step": 354
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9385023609647876e-05,
+      "loss": 0.427,
+      "step": 355
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.937951873975871e-05,
+      "loss": 0.4403,
+      "step": 356
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.937398965108074e-05,
+      "loss": 0.391,
+      "step": 357
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9368436349106606e-05,
+      "loss": 0.4068,
+      "step": 358
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9362858839352994e-05,
+      "loss": 0.3785,
+      "step": 359
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.935725712736064e-05,
+      "loss": 0.4255,
+      "step": 360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.935163121869433e-05,
+      "loss": 0.4177,
+      "step": 361
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9345981118942866e-05,
+      "loss": 0.448,
+      "step": 362
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9340306833719116e-05,
+      "loss": 0.385,
+      "step": 363
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.933460836865996e-05,
+      "loss": 0.456,
+      "step": 364
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.932888572942627e-05,
+      "loss": 0.4268,
+      "step": 365
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.932313892170298e-05,
+      "loss": 0.4266,
+      "step": 366
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.931736795119899e-05,
+      "loss": 0.4276,
+      "step": 367
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.931157282364724e-05,
+      "loss": 0.4119,
+      "step": 368
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9305753544804656e-05,
+      "loss": 0.4336,
+      "step": 369
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.929991012045213e-05,
+      "loss": 0.4006,
+      "step": 370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.929404255639457e-05,
+      "loss": 0.4335,
+      "step": 371
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.928815085846087e-05,
+      "loss": 0.41,
+      "step": 372
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.928223503250386e-05,
+      "loss": 0.3899,
+      "step": 373
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.927629508440039e-05,
+      "loss": 0.4216,
+      "step": 374
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.927033102005123e-05,
+      "loss": 0.4109,
+      "step": 375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.926434284538113e-05,
+      "loss": 0.4534,
+      "step": 376
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.925833056633878e-05,
+      "loss": 0.4261,
+      "step": 377
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.925229418889683e-05,
+      "loss": 0.4508,
+      "step": 378
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.924623371905185e-05,
+      "loss": 0.4512,
+      "step": 379
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9240149162824375e-05,
+      "loss": 0.4207,
+      "step": 380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.923404052625882e-05,
+      "loss": 0.3877,
+      "step": 381
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.922790781542357e-05,
+      "loss": 0.4854,
+      "step": 382
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9221751036410886e-05,
+      "loss": 0.4227,
+      "step": 383
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9215570195336966e-05,
+      "loss": 0.4381,
+      "step": 384
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9209365298341905e-05,
+      "loss": 0.4049,
+      "step": 385
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.920313635158969e-05,
+      "loss": 0.4068,
+      "step": 386
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9196883361268195e-05,
+      "loss": 0.4098,
+      "step": 387
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9190606333589194e-05,
+      "loss": 0.3989,
+      "step": 388
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9184305274788326e-05,
+      "loss": 0.4458,
+      "step": 389
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.917798019112511e-05,
+      "loss": 0.4216,
+      "step": 390
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.917163108888293e-05,
+      "loss": 0.4183,
+      "step": 391
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.916525797436903e-05,
+      "loss": 0.4,
+      "step": 392
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.91588608539145e-05,
+      "loss": 0.4203,
+      "step": 393
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.91524397338743e-05,
+      "loss": 0.4413,
+      "step": 394
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.91459946206272e-05,
+      "loss": 0.3917,
+      "step": 395
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.913952552057583e-05,
+      "loss": 0.4534,
+      "step": 396
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.913303244014665e-05,
+      "loss": 0.4205,
+      "step": 397
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9126515385789917e-05,
+      "loss": 0.3863,
+      "step": 398
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.911997436397973e-05,
+      "loss": 0.4193,
+      "step": 399
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.911340938121398e-05,
+      "loss": 0.382,
+      "step": 400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.910682044401437e-05,
+      "loss": 0.4107,
+      "step": 401
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.91002075589264e-05,
+      "loss": 0.4703,
+      "step": 402
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9093570732519356e-05,
+      "loss": 0.4107,
+      "step": 403
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9086909971386305e-05,
+      "loss": 0.4217,
+      "step": 404
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.90802252821441e-05,
+      "loss": 0.438,
+      "step": 405
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.907351667143336e-05,
+      "loss": 0.4488,
+      "step": 406
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9066784145918466e-05,
+      "loss": 0.4026,
+      "step": 407
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.906002771228755e-05,
+      "loss": 0.4642,
+      "step": 408
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.905324737725251e-05,
+      "loss": 0.4261,
+      "step": 409
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.904644314754898e-05,
+      "loss": 0.3915,
+      "step": 410
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9039615029936325e-05,
+      "loss": 0.4409,
+      "step": 411
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.903276303119765e-05,
+      "loss": 0.4127,
+      "step": 412
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.902588715813978e-05,
+      "loss": 0.4104,
+      "step": 413
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.901898741759325e-05,
+      "loss": 0.4067,
+      "step": 414
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.901206381641232e-05,
+      "loss": 0.4054,
+      "step": 415
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.900511636147494e-05,
+      "loss": 0.4625,
+      "step": 416
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.899814505968277e-05,
+      "loss": 0.3842,
+      "step": 417
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.899114991796114e-05,
+      "loss": 0.4408,
+      "step": 418
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8984130943259074e-05,
+      "loss": 0.4246,
+      "step": 419
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8977088142549285e-05,
+      "loss": 0.414,
+      "step": 420
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.897002152282813e-05,
+      "loss": 0.4113,
+      "step": 421
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.896293109111564e-05,
+      "loss": 0.4643,
+      "step": 422
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8955816854455504e-05,
+      "loss": 0.4039,
+      "step": 423
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.894867881991506e-05,
+      "loss": 0.4179,
+      "step": 424
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8941516994585276e-05,
+      "loss": 0.4295,
+      "step": 425
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.893433138558077e-05,
+      "loss": 0.4267,
+      "step": 426
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.892712200003976e-05,
+      "loss": 0.4365,
+      "step": 427
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.891988884512411e-05,
+      "loss": 0.4635,
+      "step": 428
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8912631928019293e-05,
+      "loss": 0.4099,
+      "step": 429
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.890535125593439e-05,
+      "loss": 0.4246,
+      "step": 430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.889804683610204e-05,
+      "loss": 0.4143,
+      "step": 431
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8890718675778544e-05,
+      "loss": 0.4049,
+      "step": 432
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.888336678224372e-05,
+      "loss": 0.4429,
+      "step": 433
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.887599116280101e-05,
+      "loss": 0.3648,
+      "step": 434
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.886859182477742e-05,
+      "loss": 0.4217,
+      "step": 435
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.886116877552347e-05,
+      "loss": 0.4278,
+      "step": 436
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.885372202241329e-05,
+      "loss": 0.4188,
+      "step": 437
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.884625157284454e-05,
+      "loss": 0.4413,
+      "step": 438
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8838757434238406e-05,
+      "loss": 0.4109,
+      "step": 439
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8831239614039636e-05,
+      "loss": 0.4332,
+      "step": 440
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.882369811971648e-05,
+      "loss": 0.3771,
+      "step": 441
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8816132958760695e-05,
+      "loss": 0.4372,
+      "step": 442
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.88085441386876e-05,
+      "loss": 0.4561,
+      "step": 443
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8800931667035946e-05,
+      "loss": 0.3918,
+      "step": 444
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8793295551368046e-05,
+      "loss": 0.4147,
+      "step": 445
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8785635799269666e-05,
+      "loss": 0.4384,
+      "step": 446
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.877795241835005e-05,
+      "loss": 0.4382,
+      "step": 447
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.8770245416241923e-05,
+      "loss": 0.4616,
+      "step": 448
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.8762514800601485e-05,
+      "loss": 0.4011,
+      "step": 449
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.875476057910838e-05,
+      "loss": 0.3968,
+      "step": 450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.8746982759465716e-05,
+      "loss": 0.3583,
+      "step": 451
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.873918134940002e-05,
+      "loss": 0.389,
+      "step": 452
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.873135635666128e-05,
+      "loss": 0.4186,
+      "step": 453
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.8723507789022894e-05,
+      "loss": 0.4022,
+      "step": 454
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.871563565428169e-05,
+      "loss": 0.4142,
+      "step": 455
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.870773996025789e-05,
+      "loss": 0.4188,
+      "step": 456
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.869982071479516e-05,
+      "loss": 0.4549,
+      "step": 457
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.869187792576051e-05,
+      "loss": 0.4231,
+      "step": 458
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.8683911601044374e-05,
+      "loss": 0.422,
+      "step": 459
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.8675921748560546e-05,
+      "loss": 0.4289,
+      "step": 460
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.8667908376246225e-05,
+      "loss": 0.4242,
+      "step": 461
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.8659871492061936e-05,
+      "loss": 0.3913,
+      "step": 462
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.8651811103991574e-05,
+      "loss": 0.3936,
+      "step": 463
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.86437272200424e-05,
+      "loss": 0.3864,
+      "step": 464
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.863561984824499e-05,
+      "loss": 0.3866,
+      "step": 465
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.862748899665327e-05,
+      "loss": 0.4266,
+      "step": 466
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.861933467334449e-05,
+      "loss": 0.3964,
+      "step": 467
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.8611156886419206e-05,
+      "loss": 0.3931,
+      "step": 468
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.8602955644001304e-05,
+      "loss": 0.4122,
+      "step": 469
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.859473095423793e-05,
+      "loss": 0.4186,
+      "step": 470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.858648282529958e-05,
+      "loss": 0.4473,
+      "step": 471
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.857821126537999e-05,
+      "loss": 0.4293,
+      "step": 472
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.856991628269619e-05,
+      "loss": 0.4055,
+      "step": 473
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.8561597885488464e-05,
+      "loss": 0.4293,
+      "step": 474
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.855325608202038e-05,
+      "loss": 0.429,
+      "step": 475
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.8544890880578755e-05,
+      "loss": 0.4079,
+      "step": 476
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.8536502289473615e-05,
+      "loss": 0.3836,
+      "step": 477
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.852809031703826e-05,
+      "loss": 0.3758,
+      "step": 478
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.851965497162921e-05,
+      "loss": 0.4545,
+      "step": 479
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.8511196261626176e-05,
+      "loss": 0.4178,
+      "step": 480
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.850271419543213e-05,
+      "loss": 0.4669,
+      "step": 481
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.84942087814732e-05,
+      "loss": 0.419,
+      "step": 482
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.848568002819873e-05,
+      "loss": 0.4498,
+      "step": 483
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.847712794408124e-05,
+      "loss": 0.4329,
+      "step": 484
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.846855253761644e-05,
+      "loss": 0.3889,
+      "step": 485
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.845995381732319e-05,
+      "loss": 0.385,
+      "step": 486
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.845133179174353e-05,
+      "loss": 0.3938,
+      "step": 487
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8442686469442625e-05,
+      "loss": 0.4303,
+      "step": 488
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.843401785900883e-05,
+      "loss": 0.3815,
+      "step": 489
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.842532596905357e-05,
+      "loss": 0.3843,
+      "step": 490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.841661080821146e-05,
+      "loss": 0.4272,
+      "step": 491
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8407872385140194e-05,
+      "loss": 0.442,
+      "step": 492
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.839911070852058e-05,
+      "loss": 0.4394,
+      "step": 493
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.839032578705652e-05,
+      "loss": 0.3777,
+      "step": 494
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8381517629475046e-05,
+      "loss": 0.3979,
+      "step": 495
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.837268624452622e-05,
+      "loss": 0.4165,
+      "step": 496
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.836383164098322e-05,
+      "loss": 0.3994,
+      "step": 497
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.835495382764226e-05,
+      "loss": 0.4031,
+      "step": 498
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.834605281332263e-05,
+      "loss": 0.3697,
+      "step": 499
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.833712860686666e-05,
+      "loss": 0.4268,
+      "step": 500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8328181217139726e-05,
+      "loss": 0.4079,
+      "step": 501
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.831921065303021e-05,
+      "loss": 0.4152,
+      "step": 502
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.831021692344956e-05,
+      "loss": 0.4513,
+      "step": 503
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8301200037332195e-05,
+      "loss": 0.4299,
+      "step": 504
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.829216000363557e-05,
+      "loss": 0.4094,
+      "step": 505
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.828309683134009e-05,
+      "loss": 0.3703,
+      "step": 506
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.827401052944921e-05,
+      "loss": 0.3849,
+      "step": 507
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.82649011069893e-05,
+      "loss": 0.4373,
+      "step": 508
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.825576857300975e-05,
+      "loss": 0.4247,
+      "step": 509
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.824661293658287e-05,
+      "loss": 0.4008,
+      "step": 510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.823743420680393e-05,
+      "loss": 0.4219,
+      "step": 511
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.822823239279115e-05,
+      "loss": 0.3738,
+      "step": 512
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.821900750368569e-05,
+      "loss": 0.3974,
+      "step": 513
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8209759548651614e-05,
+      "loss": 0.3795,
+      "step": 514
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.820048853687589e-05,
+      "loss": 0.4663,
+      "step": 515
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8191194477568435e-05,
+      "loss": 0.4167,
+      "step": 516
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8181877379962016e-05,
+      "loss": 0.4033,
+      "step": 517
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.817253725331231e-05,
+      "loss": 0.4033,
+      "step": 518
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.816317410689787e-05,
+      "loss": 0.4511,
+      "step": 519
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.815378795002011e-05,
+      "loss": 0.4129,
+      "step": 520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.81443787920033e-05,
+      "loss": 0.4043,
+      "step": 521
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.813494664219458e-05,
+      "loss": 0.3836,
+      "step": 522
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.812549150996389e-05,
+      "loss": 0.4284,
+      "step": 523
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8116013404704055e-05,
+      "loss": 0.4203,
+      "step": 524
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8106512335830665e-05,
+      "loss": 0.4202,
+      "step": 525
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8096988312782174e-05,
+      "loss": 0.4309,
+      "step": 526
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8087441345019805e-05,
+      "loss": 0.4403,
+      "step": 527
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.807787144202757e-05,
+      "loss": 0.4168,
+      "step": 528
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8068278613312307e-05,
+      "loss": 0.3675,
+      "step": 529
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.805866286840358e-05,
+      "loss": 0.4278,
+      "step": 530
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.804902421685375e-05,
+      "loss": 0.3856,
+      "step": 531
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.803936266823792e-05,
+      "loss": 0.4243,
+      "step": 532
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8029678232153935e-05,
+      "loss": 0.4421,
+      "step": 533
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.801997091822239e-05,
+      "loss": 0.439,
+      "step": 534
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.80102407360866e-05,
+      "loss": 0.4075,
+      "step": 535
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.80004876954126e-05,
+      "loss": 0.417,
+      "step": 536
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.799071180588912e-05,
+      "loss": 0.3874,
+      "step": 537
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.798091307722761e-05,
+      "loss": 0.4145,
+      "step": 538
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.7971091519162195e-05,
+      "loss": 0.4009,
+      "step": 539
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.796124714144967e-05,
+      "loss": 0.4414,
+      "step": 540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.795137995386953e-05,
+      "loss": 0.3819,
+      "step": 541
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.79414899662239e-05,
+      "loss": 0.4355,
+      "step": 542
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.7931577188337556e-05,
+      "loss": 0.4214,
+      "step": 543
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.7921641630057934e-05,
+      "loss": 0.4434,
+      "step": 544
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.791168330125509e-05,
+      "loss": 0.4291,
+      "step": 545
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.79017022118217e-05,
+      "loss": 0.4027,
+      "step": 546
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.789169837167303e-05,
+      "loss": 0.4448,
+      "step": 547
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.7881671790747e-05,
+      "loss": 0.4202,
+      "step": 548
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.7871622479004055e-05,
+      "loss": 0.4519,
+      "step": 549
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.786155044642727e-05,
+      "loss": 0.4008,
+      "step": 550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.7851455703022274e-05,
+      "loss": 0.3539,
+      "step": 551
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.7841338258817256e-05,
+      "loss": 0.4495,
+      "step": 552
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.783119812386296e-05,
+      "loss": 0.4709,
+      "step": 553
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.7821035308232665e-05,
+      "loss": 0.3633,
+      "step": 554
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.781084982202218e-05,
+      "loss": 0.4289,
+      "step": 555
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.780064167534985e-05,
+      "loss": 0.4277,
+      "step": 556
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.779041087835652e-05,
+      "loss": 0.4157,
+      "step": 557
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.778015744120553e-05,
+      "loss": 0.4221,
+      "step": 558
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.776988137408272e-05,
+      "loss": 0.3797,
+      "step": 559
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7759582687196414e-05,
+      "loss": 0.4295,
+      "step": 560
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.77492613907774e-05,
+      "loss": 0.3901,
+      "step": 561
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.773891749507893e-05,
+      "loss": 0.3927,
+      "step": 562
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.772855101037669e-05,
+      "loss": 0.3983,
+      "step": 563
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7718161946968835e-05,
+      "loss": 0.4242,
+      "step": 564
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.770775031517594e-05,
+      "loss": 0.4273,
+      "step": 565
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.769731612534097e-05,
+      "loss": 0.4001,
+      "step": 566
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.768685938782935e-05,
+      "loss": 0.3542,
+      "step": 567
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7676380113028864e-05,
+      "loss": 0.4339,
+      "step": 568
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.76658783113497e-05,
+      "loss": 0.3986,
+      "step": 569
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.765535399322443e-05,
+      "loss": 0.4115,
+      "step": 570
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.764480716910799e-05,
+      "loss": 0.416,
+      "step": 571
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.763423784947765e-05,
+      "loss": 0.3928,
+      "step": 572
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.762364604483306e-05,
+      "loss": 0.439,
+      "step": 573
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.7613031765696206e-05,
+      "loss": 0.3991,
+      "step": 574
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.760239502261138e-05,
+      "loss": 0.3757,
+      "step": 575
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.759173582614519e-05,
+      "loss": 0.3908,
+      "step": 576
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.7581054186886576e-05,
+      "loss": 0.3565,
+      "step": 577
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.757035011544674e-05,
+      "loss": 0.4322,
+      "step": 578
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.7559623622459184e-05,
+      "loss": 0.4558,
+      "step": 579
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.754887471857969e-05,
+      "loss": 0.4115,
+      "step": 580
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.75381034144863e-05,
+      "loss": 0.3925,
+      "step": 581
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7527309720879295e-05,
+      "loss": 0.4365,
+      "step": 582
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.751649364848121e-05,
+      "loss": 0.4647,
+      "step": 583
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.750565520803681e-05,
+      "loss": 0.4197,
+      "step": 584
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.749479441031308e-05,
+      "loss": 0.3991,
+      "step": 585
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.748391126609922e-05,
+      "loss": 0.3807,
+      "step": 586
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7473005786206615e-05,
+      "loss": 0.4303,
+      "step": 587
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.7462077981468844e-05,
+      "loss": 0.4369,
+      "step": 588
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.745112786274167e-05,
+      "loss": 0.4273,
+      "step": 589
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.7440155440903035e-05,
+      "loss": 0.4179,
+      "step": 590
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.7429160726853e-05,
+      "loss": 0.3864,
+      "step": 591
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.7418143731513805e-05,
+      "loss": 0.355,
+      "step": 592
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.740710446582981e-05,
+      "loss": 0.3731,
+      "step": 593
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.739604294076749e-05,
+      "loss": 0.4099,
+      "step": 594
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.738495916731546e-05,
+      "loss": 0.4085,
+      "step": 595
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.7373853156484406e-05,
+      "loss": 0.3578,
+      "step": 596
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.736272491930713e-05,
+      "loss": 0.3836,
+      "step": 597
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.73515744668385e-05,
+      "loss": 0.4039,
+      "step": 598
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.734040181015545e-05,
+      "loss": 0.4203,
+      "step": 599
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.7329206960356974e-05,
+      "loss": 0.3953,
+      "step": 600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.731798992856413e-05,
+      "loss": 0.4122,
+      "step": 601
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.730675072592e-05,
+      "loss": 0.4113,
+      "step": 602
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.729548936358966e-05,
+      "loss": 0.3748,
+      "step": 603
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.728420585276026e-05,
+      "loss": 0.3952,
+      "step": 604
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.727290020464091e-05,
+      "loss": 0.4183,
+      "step": 605
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.726157243046271e-05,
+      "loss": 0.3987,
+      "step": 606
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.725022254147877e-05,
+      "loss": 0.3762,
+      "step": 607
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.723885054896414e-05,
+      "loss": 0.3957,
+      "step": 608
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.7227456464215845e-05,
+      "loss": 0.4402,
+      "step": 609
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.721604029855284e-05,
+      "loss": 0.4241,
+      "step": 610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.720460206331604e-05,
+      "loss": 0.3628,
+      "step": 611
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.7193141769868265e-05,
+      "loss": 0.4692,
+      "step": 612
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.718165942959425e-05,
+      "loss": 0.4082,
+      "step": 613
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.717015505390063e-05,
+      "loss": 0.357,
+      "step": 614
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.715862865421594e-05,
+      "loss": 0.3916,
+      "step": 615
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.714708024199059e-05,
+      "loss": 0.4301,
+      "step": 616
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.713550982869685e-05,
+      "loss": 0.4037,
+      "step": 617
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.7123917425828864e-05,
+      "loss": 0.3739,
+      "step": 618
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.711230304490259e-05,
+      "loss": 0.3737,
+      "step": 619
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.7100666697455855e-05,
+      "loss": 0.4284,
+      "step": 620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.708900839504828e-05,
+      "loss": 0.4227,
+      "step": 621
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.70773281492613e-05,
+      "loss": 0.3883,
+      "step": 622
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.7065625971698166e-05,
+      "loss": 0.404,
+      "step": 623
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.7053901873983906e-05,
+      "loss": 0.3784,
+      "step": 624
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.704215586776532e-05,
+      "loss": 0.4182,
+      "step": 625
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.7030387964710964e-05,
+      "loss": 0.4029,
+      "step": 626
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.7018598176511175e-05,
+      "loss": 0.4465,
+      "step": 627
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.7006786514877997e-05,
+      "loss": 0.4264,
+      "step": 628
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.699495299154524e-05,
+      "loss": 0.382,
+      "step": 629
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.698309761826839e-05,
+      "loss": 0.4081,
+      "step": 630
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.697122040682467e-05,
+      "loss": 0.3995,
+      "step": 631
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.695932136901297e-05,
+      "loss": 0.4922,
+      "step": 632
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6947400516653916e-05,
+      "loss": 0.3655,
+      "step": 633
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.693545786158974e-05,
+      "loss": 0.4442,
+      "step": 634
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6923493415684366e-05,
+      "loss": 0.3619,
+      "step": 635
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.691150719082336e-05,
+      "loss": 0.4131,
+      "step": 636
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6899499198913935e-05,
+      "loss": 0.3937,
+      "step": 637
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6887469451884904e-05,
+      "loss": 0.4046,
+      "step": 638
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.68754179616867e-05,
+      "loss": 0.3956,
+      "step": 639
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6863344740291376e-05,
+      "loss": 0.4159,
+      "step": 640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6851249799692544e-05,
+      "loss": 0.442,
+      "step": 641
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.683913315190541e-05,
+      "loss": 0.4142,
+      "step": 642
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.682699480896674e-05,
+      "loss": 0.3834,
+      "step": 643
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6814834782934844e-05,
+      "loss": 0.3856,
+      "step": 644
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.680265308588959e-05,
+      "loss": 0.3921,
+      "step": 645
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.679044972993235e-05,
+      "loss": 0.4124,
+      "step": 646
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6778224727186036e-05,
+      "loss": 0.4089,
+      "step": 647
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.676597808979505e-05,
+      "loss": 0.3765,
+      "step": 648
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.67537098299253e-05,
+      "loss": 0.4294,
+      "step": 649
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.674141995976413e-05,
+      "loss": 0.3724,
+      "step": 650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.672910849152042e-05,
+      "loss": 0.443,
+      "step": 651
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.671677543742447e-05,
+      "loss": 0.4528,
+      "step": 652
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.670442080972799e-05,
+      "loss": 0.3716,
+      "step": 653
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.669204462070419e-05,
+      "loss": 0.3829,
+      "step": 654
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.667964688264765e-05,
+      "loss": 0.4069,
+      "step": 655
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.666722760787439e-05,
+      "loss": 0.4162,
+      "step": 656
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.665478680872177e-05,
+      "loss": 0.3745,
+      "step": 657
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6642324497548615e-05,
+      "loss": 0.3942,
+      "step": 658
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.662984068673503e-05,
+      "loss": 0.4009,
+      "step": 659
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6617335388682556e-05,
+      "loss": 0.4084,
+      "step": 660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6604808615814034e-05,
+      "loss": 0.3871,
+      "step": 661
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.659226038057364e-05,
+      "loss": 0.375,
+      "step": 662
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6579690695426894e-05,
+      "loss": 0.3837,
+      "step": 663
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.656709957286061e-05,
+      "loss": 0.4149,
+      "step": 664
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.655448702538289e-05,
+      "loss": 0.407,
+      "step": 665
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6541853065523135e-05,
+      "loss": 0.3839,
+      "step": 666
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.652919770583202e-05,
+      "loss": 0.4184,
+      "step": 667
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6516520958881455e-05,
+      "loss": 0.4045,
+      "step": 668
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.650382283726462e-05,
+      "loss": 0.4691,
+      "step": 669
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.6491103353595924e-05,
+      "loss": 0.4022,
+      "step": 670
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.6478362520510984e-05,
+      "loss": 0.4309,
+      "step": 671
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.646560035066664e-05,
+      "loss": 0.3683,
+      "step": 672
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.645281685674092e-05,
+      "loss": 0.373,
+      "step": 673
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.644001205143304e-05,
+      "loss": 0.3746,
+      "step": 674
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.6427185947463395e-05,
+      "loss": 0.3887,
+      "step": 675
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.641433855757351e-05,
+      "loss": 0.3711,
+      "step": 676
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.6401469894526086e-05,
+      "loss": 0.4816,
+      "step": 677
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.638857997110494e-05,
+      "loss": 0.4084,
+      "step": 678
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.637566880011502e-05,
+      "loss": 0.4029,
+      "step": 679
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.6362736394382377e-05,
+      "loss": 0.4012,
+      "step": 680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.634978276675415e-05,
+      "loss": 0.382,
+      "step": 681
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.633680793009857e-05,
+      "loss": 0.4043,
+      "step": 682
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.632381189730493e-05,
+      "loss": 0.4308,
+      "step": 683
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.631079468128358e-05,
+      "loss": 0.3943,
+      "step": 684
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.6297756294965924e-05,
+      "loss": 0.3682,
+      "step": 685
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.628469675130438e-05,
+      "loss": 0.3899,
+      "step": 686
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.6271616063272395e-05,
+      "loss": 0.3974,
+      "step": 687
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.6258514243864414e-05,
+      "loss": 0.3478,
+      "step": 688
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.6245391306095886e-05,
+      "loss": 0.4219,
+      "step": 689
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.623224726300324e-05,
+      "loss": 0.3915,
+      "step": 690
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.621908212764383e-05,
+      "loss": 0.4598,
+      "step": 691
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.620589591309603e-05,
+      "loss": 0.4181,
+      "step": 692
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.61926886324591e-05,
+      "loss": 0.3603,
+      "step": 693
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.617946029885327e-05,
+      "loss": 0.4266,
+      "step": 694
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.616621092541963e-05,
+      "loss": 0.4007,
+      "step": 695
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.615294052532023e-05,
+      "loss": 0.4067,
+      "step": 696
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.6139649111737966e-05,
+      "loss": 0.4099,
+      "step": 697
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.612633669787663e-05,
+      "loss": 0.3774,
+      "step": 698
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.611300329696085e-05,
+      "loss": 0.3726,
+      "step": 699
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.609964892223616e-05,
+      "loss": 0.3991,
+      "step": 700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.608627358696886e-05,
+      "loss": 0.4502,
+      "step": 701
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.607287730444612e-05,
+      "loss": 0.3525,
+      "step": 702
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.6059460087975883e-05,
+      "loss": 0.3536,
+      "step": 703
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.604602195088694e-05,
+      "loss": 0.3968,
+      "step": 704
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.6032562906528805e-05,
+      "loss": 0.3831,
+      "step": 705
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.60190829682718e-05,
+      "loss": 0.3752,
+      "step": 706
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.6005582149507e-05,
+      "loss": 0.3724,
+      "step": 707
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.59920604636462e-05,
+      "loss": 0.3918,
+      "step": 708
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5978517924121944e-05,
+      "loss": 0.3546,
+      "step": 709
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.59649545443875e-05,
+      "loss": 0.4078,
+      "step": 710
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.595137033791681e-05,
+      "loss": 0.3843,
+      "step": 711
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.593776531820453e-05,
+      "loss": 0.4055,
+      "step": 712
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.592413949876598e-05,
+      "loss": 0.4293,
+      "step": 713
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.591049289313716e-05,
+      "loss": 0.4378,
+      "step": 714
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5896825514874695e-05,
+      "loss": 0.3794,
+      "step": 715
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.588313737755585e-05,
+      "loss": 0.3798,
+      "step": 716
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.586942849477853e-05,
+      "loss": 0.4133,
+      "step": 717
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.585569888016125e-05,
+      "loss": 0.4109,
+      "step": 718
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.584194854734307e-05,
+      "loss": 0.3706,
+      "step": 719
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5828177509983704e-05,
+      "loss": 0.4253,
+      "step": 720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.581438578176338e-05,
+      "loss": 0.4375,
+      "step": 721
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.580057337638291e-05,
+      "loss": 0.3776,
+      "step": 722
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5786740307563636e-05,
+      "loss": 0.4261,
+      "step": 723
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.577288658904741e-05,
+      "loss": 0.4265,
+      "step": 724
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.575901223459665e-05,
+      "loss": 0.4174,
+      "step": 725
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5745117257994204e-05,
+      "loss": 0.3963,
+      "step": 726
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5731201673043465e-05,
+      "loss": 0.3594,
+      "step": 727
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.571726549356827e-05,
+      "loss": 0.4076,
+      "step": 728
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5703308733412905e-05,
+      "loss": 0.3558,
+      "step": 729
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.568933140644214e-05,
+      "loss": 0.3765,
+      "step": 730
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.567533352654113e-05,
+      "loss": 0.3959,
+      "step": 731
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.566131510761548e-05,
+      "loss": 0.3816,
+      "step": 732
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5647276163591193e-05,
+      "loss": 0.4148,
+      "step": 733
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5633216708414636e-05,
+      "loss": 0.4675,
+      "step": 734
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.561913675605259e-05,
+      "loss": 0.4162,
+      "step": 735
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.560503632049218e-05,
+      "loss": 0.4059,
+      "step": 736
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5590915415740886e-05,
+      "loss": 0.3818,
+      "step": 737
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5576774055826494e-05,
+      "loss": 0.4365,
+      "step": 738
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5562612254797156e-05,
+      "loss": 0.417,
+      "step": 739
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.554843002672129e-05,
+      "loss": 0.3968,
+      "step": 740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5534227385687625e-05,
+      "loss": 0.4353,
+      "step": 741
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.552000434580518e-05,
+      "loss": 0.3705,
+      "step": 742
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.550576092120321e-05,
+      "loss": 0.3872,
+      "step": 743
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.549149712603123e-05,
+      "loss": 0.4114,
+      "step": 744
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5477212974459024e-05,
+      "loss": 0.4438,
+      "step": 745
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.546290848067655e-05,
+      "loss": 0.4145,
+      "step": 746
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5448583658894e-05,
+      "loss": 0.4264,
+      "step": 747
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5434238523341756e-05,
+      "loss": 0.4013,
+      "step": 748
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.541987308827036e-05,
+      "loss": 0.4186,
+      "step": 749
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.540548736795057e-05,
+      "loss": 0.3937,
+      "step": 750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.539108137667324e-05,
+      "loss": 0.4502,
+      "step": 751
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.53766551287494e-05,
+      "loss": 0.4525,
+      "step": 752
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.536220863851018e-05,
+      "loss": 0.3497,
+      "step": 753
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.534774192030682e-05,
+      "loss": 0.411,
+      "step": 754
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5333254988510674e-05,
+      "loss": 0.3974,
+      "step": 755
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.531874785751317e-05,
+      "loss": 0.3778,
+      "step": 756
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.530422054172577e-05,
+      "loss": 0.4231,
+      "step": 757
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.528967305558004e-05,
+      "loss": 0.4027,
+      "step": 758
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.527510541352754e-05,
+      "loss": 0.3888,
+      "step": 759
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5260517630039875e-05,
+      "loss": 0.3959,
+      "step": 760
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.524590971960866e-05,
+      "loss": 0.4051,
+      "step": 761
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.52312816967455e-05,
+      "loss": 0.4242,
+      "step": 762
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.521663357598196e-05,
+      "loss": 0.3845,
+      "step": 763
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5201965371869605e-05,
+      "loss": 0.3943,
+      "step": 764
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.518727709897993e-05,
+      "loss": 0.4247,
+      "step": 765
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5172568771904365e-05,
+      "loss": 0.3656,
+      "step": 766
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.515784040525428e-05,
+      "loss": 0.4017,
+      "step": 767
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.514309201366092e-05,
+      "loss": 0.4411,
+      "step": 768
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.512832361177547e-05,
+      "loss": 0.3765,
+      "step": 769
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.511353521426894e-05,
+      "loss": 0.354,
+      "step": 770
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.509872683583225e-05,
+      "loss": 0.3832,
+      "step": 771
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5083898491176136e-05,
+      "loss": 0.3588,
+      "step": 772
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.506905019503118e-05,
+      "loss": 0.3925,
+      "step": 773
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.50541819621478e-05,
+      "loss": 0.4081,
+      "step": 774
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.50392938072962e-05,
+      "loss": 0.4115,
+      "step": 775
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5024385745266385e-05,
+      "loss": 0.3937,
+      "step": 776
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5009457790868116e-05,
+      "loss": 0.4319,
+      "step": 777
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.499450995893093e-05,
+      "loss": 0.4229,
+      "step": 778
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.497954226430413e-05,
+      "loss": 0.3221,
+      "step": 779
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.496455472185671e-05,
+      "loss": 0.4039,
+      "step": 780
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.494954734647741e-05,
+      "loss": 0.3786,
+      "step": 781
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.4934520153074686e-05,
+      "loss": 0.4583,
+      "step": 782
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.491947315657662e-05,
+      "loss": 0.4277,
+      "step": 783
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.4904406371931035e-05,
+      "loss": 0.4292,
+      "step": 784
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.488931981410537e-05,
+      "loss": 0.3815,
+      "step": 785
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.487421349808673e-05,
+      "loss": 0.4036,
+      "step": 786
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.485908743888182e-05,
+      "loss": 0.3786,
+      "step": 787
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4843941651517e-05,
+      "loss": 0.4144,
+      "step": 788
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.482877615103818e-05,
+      "loss": 0.423,
+      "step": 789
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4813590952510884e-05,
+      "loss": 0.4139,
+      "step": 790
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.479838607102021e-05,
+      "loss": 0.3907,
+      "step": 791
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.478316152167079e-05,
+      "loss": 0.4419,
+      "step": 792
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.476791731958679e-05,
+      "loss": 0.4257,
+      "step": 793
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.475265347991193e-05,
+      "loss": 0.36,
+      "step": 794
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.47373700178094e-05,
+      "loss": 0.3142,
+      "step": 795
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.47220669484619e-05,
+      "loss": 0.4051,
+      "step": 796
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.470674428707162e-05,
+      "loss": 0.3486,
+      "step": 797
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.469140204886019e-05,
+      "loss": 0.3999,
+      "step": 798
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.467604024906872e-05,
+      "loss": 0.4069,
+      "step": 799
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.466065890295771e-05,
+      "loss": 0.3953,
+      "step": 800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4645258025807104e-05,
+      "loss": 0.3736,
+      "step": 801
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.462983763291626e-05,
+      "loss": 0.4168,
+      "step": 802
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.461439773960388e-05,
+      "loss": 0.3911,
+      "step": 803
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.4598938361208095e-05,
+      "loss": 0.3976,
+      "step": 804
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.458345951308634e-05,
+      "loss": 0.4244,
+      "step": 805
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.456796121061542e-05,
+      "loss": 0.482,
+      "step": 806
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.455244346919146e-05,
+      "loss": 0.4277,
+      "step": 807
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.4536906304229894e-05,
+      "loss": 0.4032,
+      "step": 808
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.4521349731165465e-05,
+      "loss": 0.3785,
+      "step": 809
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4505773765452164e-05,
+      "loss": 0.4041,
+      "step": 810
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.449017842256328e-05,
+      "loss": 0.3898,
+      "step": 811
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.447456371799134e-05,
+      "loss": 0.4226,
+      "step": 812
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.44589296672481e-05,
+      "loss": 0.431,
+      "step": 813
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4443276285864536e-05,
+      "loss": 0.3526,
+      "step": 814
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.442760358939083e-05,
+      "loss": 0.4172,
+      "step": 815
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.441191159339636e-05,
+      "loss": 0.3732,
+      "step": 816
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.439620031346965e-05,
+      "loss": 0.4203,
+      "step": 817
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4380469765218416e-05,
+      "loss": 0.408,
+      "step": 818
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.436471996426949e-05,
+      "loss": 0.3849,
+      "step": 819
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.434895092626883e-05,
+      "loss": 0.4039,
+      "step": 820
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4333162666881515e-05,
+      "loss": 0.3859,
+      "step": 821
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.6144305467605591,
+      "eval_runtime": 144.5541,
+      "eval_samples_per_second": 5.271,
+      "eval_steps_per_second": 0.332,
+      "step": 821
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.431735520179173e-05,
+      "loss": 0.365,
+      "step": 822
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4301528546702704e-05,
+      "loss": 0.2937,
+      "step": 823
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4285682717336766e-05,
+      "loss": 0.2966,
+      "step": 824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.426981772943526e-05,
+      "loss": 0.2868,
+      "step": 825
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.425393359875859e-05,
+      "loss": 0.2714,
+      "step": 826
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.423803034108617e-05,
+      "loss": 0.2791,
+      "step": 827
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.42221079722164e-05,
+      "loss": 0.2639,
+      "step": 828
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.420616650796667e-05,
+      "loss": 0.2966,
+      "step": 829
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.419020596417336e-05,
+      "loss": 0.2724,
+      "step": 830
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.417422635669178e-05,
+      "loss": 0.2734,
+      "step": 831
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.415822770139619e-05,
+      "loss": 0.2573,
+      "step": 832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.414221001417977e-05,
+      "loss": 0.3003,
+      "step": 833
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.412617331095461e-05,
+      "loss": 0.2987,
+      "step": 834
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.411011760765167e-05,
+      "loss": 0.2872,
+      "step": 835
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.409404292022081e-05,
+      "loss": 0.2599,
+      "step": 836
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4077949264630746e-05,
+      "loss": 0.2881,
+      "step": 837
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.406183665686903e-05,
+      "loss": 0.2592,
+      "step": 838
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.404570511294203e-05,
+      "loss": 0.251,
+      "step": 839
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.402955464887494e-05,
+      "loss": 0.2388,
+      "step": 840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.401338528071175e-05,
+      "loss": 0.2789,
+      "step": 841
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.399719702451523e-05,
+      "loss": 0.2746,
+      "step": 842
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.398098989636691e-05,
+      "loss": 0.2697,
+      "step": 843
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.396476391236707e-05,
+      "loss": 0.2604,
+      "step": 844
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.3948519088634706e-05,
+      "loss": 0.2473,
+      "step": 845
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.393225544130756e-05,
+      "loss": 0.2835,
+      "step": 846
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.391597298654204e-05,
+      "loss": 0.2447,
+      "step": 847
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.389967174051326e-05,
+      "loss": 0.283,
+      "step": 848
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.3883351719414986e-05,
+      "loss": 0.251,
+      "step": 849
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.386701293945966e-05,
+      "loss": 0.2803,
+      "step": 850
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.385065541687834e-05,
+      "loss": 0.2466,
+      "step": 851
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.38342791679207e-05,
+      "loss": 0.2515,
+      "step": 852
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.381788420885502e-05,
+      "loss": 0.2755,
+      "step": 853
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.3801470555968176e-05,
+      "loss": 0.284,
+      "step": 854
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.378503822556561e-05,
+      "loss": 0.2779,
+      "step": 855
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.3768587233971314e-05,
+      "loss": 0.2758,
+      "step": 856
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.375211759752782e-05,
+      "loss": 0.2599,
+      "step": 857
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.373562933259618e-05,
+      "loss": 0.2558,
+      "step": 858
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.371912245555597e-05,
+      "loss": 0.2527,
+      "step": 859
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.370259698280522e-05,
+      "loss": 0.3286,
+      "step": 860
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.368605293076047e-05,
+      "loss": 0.2683,
+      "step": 861
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.366949031585669e-05,
+      "loss": 0.2141,
+      "step": 862
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.3652909154547295e-05,
+      "loss": 0.2762,
+      "step": 863
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.363630946330415e-05,
+      "loss": 0.2504,
+      "step": 864
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.361969125861748e-05,
+      "loss": 0.2504,
+      "step": 865
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.360305455699596e-05,
+      "loss": 0.2571,
+      "step": 866
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.358639937496658e-05,
+      "loss": 0.271,
+      "step": 867
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.356972572907473e-05,
+      "loss": 0.2745,
+      "step": 868
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.355303363588414e-05,
+      "loss": 0.2642,
+      "step": 869
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.353632311197683e-05,
+      "loss": 0.2801,
+      "step": 870
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.3519594173953184e-05,
+      "loss": 0.2641,
+      "step": 871
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.350284683843182e-05,
+      "loss": 0.307,
+      "step": 872
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.348608112204968e-05,
+      "loss": 0.2768,
+      "step": 873
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.346929704146194e-05,
+      "loss": 0.2472,
+      "step": 874
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.345249461334203e-05,
+      "loss": 0.2615,
+      "step": 875
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.3435673854381596e-05,
+      "loss": 0.2858,
+      "step": 876
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.341883478129051e-05,
+      "loss": 0.2157,
+      "step": 877
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.3401977410796816e-05,
+      "loss": 0.2644,
+      "step": 878
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.338510175964676e-05,
+      "loss": 0.301,
+      "step": 879
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.336820784460472e-05,
+      "loss": 0.2866,
+      "step": 880
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.335129568245326e-05,
+      "loss": 0.2569,
+      "step": 881
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.3334365289993006e-05,
+      "loss": 0.2894,
+      "step": 882
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.331741668404276e-05,
+      "loss": 0.2871,
+      "step": 883
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.3300449881439375e-05,
+      "loss": 0.2468,
+      "step": 884
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.32834648990378e-05,
+      "loss": 0.2801,
+      "step": 885
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.326646175371103e-05,
+      "loss": 0.2528,
+      "step": 886
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.3249440462350125e-05,
+      "loss": 0.2772,
+      "step": 887
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.323240104186414e-05,
+      "loss": 0.2506,
+      "step": 888
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.321534350918017e-05,
+      "loss": 0.254,
+      "step": 889
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.319826788124329e-05,
+      "loss": 0.2608,
+      "step": 890
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.318117417501656e-05,
+      "loss": 0.3031,
+      "step": 891
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.316406240748097e-05,
+      "loss": 0.2836,
+      "step": 892
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.3146932595635486e-05,
+      "loss": 0.2478,
+      "step": 893
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.312978475649698e-05,
+      "loss": 0.2619,
+      "step": 894
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.311261890710024e-05,
+      "loss": 0.279,
+      "step": 895
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.309543506449795e-05,
+      "loss": 0.2794,
+      "step": 896
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.307823324576066e-05,
+      "loss": 0.2921,
+      "step": 897
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.3061013467976774e-05,
+      "loss": 0.2988,
+      "step": 898
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.304377574825255e-05,
+      "loss": 0.2785,
+      "step": 899
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.302652010371205e-05,
+      "loss": 0.2895,
+      "step": 900
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.300924655149717e-05,
+      "loss": 0.2582,
+      "step": 901
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.299195510876757e-05,
+      "loss": 0.274,
+      "step": 902
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.29746457927007e-05,
+      "loss": 0.278,
+      "step": 903
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.295731862049176e-05,
+      "loss": 0.2648,
+      "step": 904
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.293997360935368e-05,
+      "loss": 0.3118,
+      "step": 905
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.2922610776517124e-05,
+      "loss": 0.28,
+      "step": 906
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.290523013923045e-05,
+      "loss": 0.2737,
+      "step": 907
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.288783171475972e-05,
+      "loss": 0.2802,
+      "step": 908
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.287041552038863e-05,
+      "loss": 0.2543,
+      "step": 909
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.285298157341858e-05,
+      "loss": 0.2927,
+      "step": 910
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.283552989116856e-05,
+      "loss": 0.2742,
+      "step": 911
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.28180604909752e-05,
+      "loss": 0.3007,
+      "step": 912
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.280057339019273e-05,
+      "loss": 0.2612,
+      "step": 913
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.278306860619296e-05,
+      "loss": 0.3012,
+      "step": 914
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.2765546156365265e-05,
+      "loss": 0.2961,
+      "step": 915
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.274800605811658e-05,
+      "loss": 0.2817,
+      "step": 916
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.273044832887136e-05,
+      "loss": 0.2539,
+      "step": 917
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.2712872986071575e-05,
+      "loss": 0.2991,
+      "step": 918
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.26952800471767e-05,
+      "loss": 0.2627,
+      "step": 919
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.267766952966369e-05,
+      "loss": 0.2926,
+      "step": 920
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.266004145102695e-05,
+      "loss": 0.2556,
+      "step": 921
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.264239582877836e-05,
+      "loss": 0.267,
+      "step": 922
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.262473268044719e-05,
+      "loss": 0.2678,
+      "step": 923
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.260705202358014e-05,
+      "loss": 0.2712,
+      "step": 924
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.25893538757413e-05,
+      "loss": 0.2692,
+      "step": 925
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.2571638254512145e-05,
+      "loss": 0.258,
+      "step": 926
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.255390517749149e-05,
+      "loss": 0.2723,
+      "step": 927
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.253615466229551e-05,
+      "loss": 0.2883,
+      "step": 928
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.25183867265577e-05,
+      "loss": 0.3127,
+      "step": 929
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.250060138792884e-05,
+      "loss": 0.2695,
+      "step": 930
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.2482798664077016e-05,
+      "loss": 0.2622,
+      "step": 931
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.246497857268759e-05,
+      "loss": 0.2745,
+      "step": 932
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.244714113146317e-05,
+      "loss": 0.2897,
+      "step": 933
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.242928635812358e-05,
+      "loss": 0.2892,
+      "step": 934
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.241141427040588e-05,
+      "loss": 0.2626,
+      "step": 935
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.239352488606436e-05,
+      "loss": 0.2549,
+      "step": 936
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.237561822287044e-05,
+      "loss": 0.2646,
+      "step": 937
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.2357694298612724e-05,
+      "loss": 0.2554,
+      "step": 938
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.233975313109698e-05,
+      "loss": 0.3118,
+      "step": 939
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.232179473814609e-05,
+      "loss": 0.2609,
+      "step": 940
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.230381913760004e-05,
+      "loss": 0.2368,
+      "step": 941
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.228582634731594e-05,
+      "loss": 0.2779,
+      "step": 942
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.226781638516794e-05,
+      "loss": 0.3054,
+      "step": 943
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.224978926904726e-05,
+      "loss": 0.2897,
+      "step": 944
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.223174501686219e-05,
+      "loss": 0.2529,
+      "step": 945
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.2213683646538005e-05,
+      "loss": 0.2947,
+      "step": 946
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.2195605176017e-05,
+      "loss": 0.3047,
+      "step": 947
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.2177509623258456e-05,
+      "loss": 0.27,
+      "step": 948
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.215939700623863e-05,
+      "loss": 0.2555,
+      "step": 949
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.214126734295072e-05,
+      "loss": 0.2863,
+      "step": 950
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.2123120651404865e-05,
+      "loss": 0.2993,
+      "step": 951
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.210495694962813e-05,
+      "loss": 0.2803,
+      "step": 952
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.208677625566445e-05,
+      "loss": 0.3022,
+      "step": 953
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.206857858757467e-05,
+      "loss": 0.2633,
+      "step": 954
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.205036396343647e-05,
+      "loss": 0.2567,
+      "step": 955
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.2032132401344405e-05,
+      "loss": 0.2843,
+      "step": 956
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.201388391940983e-05,
+      "loss": 0.273,
+      "step": 957
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.1995618535760907e-05,
+      "loss": 0.3013,
+      "step": 958
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.197733626854262e-05,
+      "loss": 0.2947,
+      "step": 959
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.195903713591669e-05,
+      "loss": 0.2874,
+      "step": 960
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.1940721156061627e-05,
+      "loss": 0.288,
+      "step": 961
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.1922388347172626e-05,
+      "loss": 0.2746,
+      "step": 962
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.1904038727461644e-05,
+      "loss": 0.2915,
+      "step": 963
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.1885672315157346e-05,
+      "loss": 0.257,
+      "step": 964
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.1867289128505016e-05,
+      "loss": 0.276,
+      "step": 965
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.184888918576668e-05,
+      "loss": 0.2761,
+      "step": 966
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.183047250522096e-05,
+      "loss": 0.2637,
+      "step": 967
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.181203910516312e-05,
+      "loss": 0.2617,
+      "step": 968
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.179358900390502e-05,
+      "loss": 0.2787,
+      "step": 969
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.1775122219775146e-05,
+      "loss": 0.2787,
+      "step": 970
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.1756638771118504e-05,
+      "loss": 0.2664,
+      "step": 971
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.1738138676296716e-05,
+      "loss": 0.3246,
+      "step": 972
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.171962195368789e-05,
+      "loss": 0.2646,
+      "step": 973
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.170108862168668e-05,
+      "loss": 0.2954,
+      "step": 974
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.168253869870422e-05,
+      "loss": 0.2871,
+      "step": 975
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.166397220316815e-05,
+      "loss": 0.282,
+      "step": 976
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.164538915352254e-05,
+      "loss": 0.2673,
+      "step": 977
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.162678956822794e-05,
+      "loss": 0.2685,
+      "step": 978
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.160817346576131e-05,
+      "loss": 0.2884,
+      "step": 979
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.1589540864616025e-05,
+      "loss": 0.2556,
+      "step": 980
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.1570891783301843e-05,
+      "loss": 0.2735,
+      "step": 981
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.155222624034488e-05,
+      "loss": 0.2791,
+      "step": 982
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.153354425428764e-05,
+      "loss": 0.3444,
+      "step": 983
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.151484584368892e-05,
+      "loss": 0.2992,
+      "step": 984
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.1496131027123874e-05,
+      "loss": 0.2849,
+      "step": 985
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.147739982318393e-05,
+      "loss": 0.2708,
+      "step": 986
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.14586522504768e-05,
+      "loss": 0.2635,
+      "step": 987
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.143988832762646e-05,
+      "loss": 0.2631,
+      "step": 988
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.142110807327312e-05,
+      "loss": 0.2577,
+      "step": 989
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.1402311506073216e-05,
+      "loss": 0.2772,
+      "step": 990
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.1383498644699404e-05,
+      "loss": 0.2861,
+      "step": 991
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.13646695078405e-05,
+      "loss": 0.3033,
+      "step": 992
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.134582411420153e-05,
+      "loss": 0.2652,
+      "step": 993
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.132696248250362e-05,
+      "loss": 0.2726,
+      "step": 994
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.130808463148405e-05,
+      "loss": 0.2733,
+      "step": 995
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.128919057989622e-05,
+      "loss": 0.2649,
+      "step": 996
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.127028034650963e-05,
+      "loss": 0.2642,
+      "step": 997
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.1251353950109804e-05,
+      "loss": 0.2713,
+      "step": 998
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.12324114094984e-05,
+      "loss": 0.3142,
+      "step": 999
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.121345274349305e-05,
+      "loss": 0.2788,
+      "step": 1000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.119447797092745e-05,
+      "loss": 0.2852,
+      "step": 1001
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.1175487110651245e-05,
+      "loss": 0.2591,
+      "step": 1002
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.115648018153011e-05,
+      "loss": 0.2746,
+      "step": 1003
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.1137457202445665e-05,
+      "loss": 0.291,
+      "step": 1004
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.111841819229547e-05,
+      "loss": 0.2594,
+      "step": 1005
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.109936316999302e-05,
+      "loss": 0.265,
+      "step": 1006
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.1080292154467705e-05,
+      "loss": 0.26,
+      "step": 1007
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.106120516466481e-05,
+      "loss": 0.272,
+      "step": 1008
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.104210221954549e-05,
+      "loss": 0.2469,
+      "step": 1009
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.102298333808674e-05,
+      "loss": 0.251,
+      "step": 1010
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.10038485392814e-05,
+      "loss": 0.3106,
+      "step": 1011
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.098469784213812e-05,
+      "loss": 0.2995,
+      "step": 1012
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.0965531265681334e-05,
+      "loss": 0.2517,
+      "step": 1013
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.0946348828951256e-05,
+      "loss": 0.232,
+      "step": 1014
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.092715055100387e-05,
+      "loss": 0.2846,
+      "step": 1015
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.0907936450910854e-05,
+      "loss": 0.272,
+      "step": 1016
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.088870654775966e-05,
+      "loss": 0.2853,
+      "step": 1017
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.086946086065341e-05,
+      "loss": 0.2616,
+      "step": 1018
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.085019940871089e-05,
+      "loss": 0.3164,
+      "step": 1019
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.0830922211066566e-05,
+      "loss": 0.2782,
+      "step": 1020
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.081162928687057e-05,
+      "loss": 0.2558,
+      "step": 1021
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.07923206552886e-05,
+      "loss": 0.303,
+      "step": 1022
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.077299633550202e-05,
+      "loss": 0.2849,
+      "step": 1023
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.075365634670771e-05,
+      "loss": 0.2605,
+      "step": 1024
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.073430070811819e-05,
+      "loss": 0.2884,
+      "step": 1025
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.071492943896147e-05,
+      "loss": 0.2757,
+      "step": 1026
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.0695542558481116e-05,
+      "loss": 0.2639,
+      "step": 1027
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.0676140085936186e-05,
+      "loss": 0.2535,
+      "step": 1028
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.065672204060125e-05,
+      "loss": 0.2611,
+      "step": 1029
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.063728844176631e-05,
+      "loss": 0.2741,
+      "step": 1030
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.0617839308736864e-05,
+      "loss": 0.3098,
+      "step": 1031
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.059837466083382e-05,
+      "loss": 0.2865,
+      "step": 1032
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.0578894517393476e-05,
+      "loss": 0.274,
+      "step": 1033
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.055939889776756e-05,
+      "loss": 0.2737,
+      "step": 1034
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.053988782132318e-05,
+      "loss": 0.2647,
+      "step": 1035
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.052036130744275e-05,
+      "loss": 0.3074,
+      "step": 1036
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.050081937552407e-05,
+      "loss": 0.2934,
+      "step": 1037
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.048126204498022e-05,
+      "loss": 0.2571,
+      "step": 1038
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.04616893352396e-05,
+      "loss": 0.2664,
+      "step": 1039
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.044210126574586e-05,
+      "loss": 0.3062,
+      "step": 1040
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.0422497855957964e-05,
+      "loss": 0.2816,
+      "step": 1041
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.040287912535006e-05,
+      "loss": 0.2386,
+      "step": 1042
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.0383245093411545e-05,
+      "loss": 0.2869,
+      "step": 1043
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.0363595779647e-05,
+      "loss": 0.2776,
+      "step": 1044
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.0343931203576186e-05,
+      "loss": 0.2728,
+      "step": 1045
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.032425138473405e-05,
+      "loss": 0.2884,
+      "step": 1046
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.0304556342670655e-05,
+      "loss": 0.2721,
+      "step": 1047
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.02848460969512e-05,
+      "loss": 0.2831,
+      "step": 1048
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.0265120667155983e-05,
+      "loss": 0.2714,
+      "step": 1049
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.02453800728804e-05,
+      "loss": 0.2493,
+      "step": 1050
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.022562433373488e-05,
+      "loss": 0.2858,
+      "step": 1051
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.0205853469344926e-05,
+      "loss": 0.2826,
+      "step": 1052
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.018606749935106e-05,
+      "loss": 0.354,
+      "step": 1053
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.0166266443408794e-05,
+      "loss": 0.2871,
+      "step": 1054
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.014645032118864e-05,
+      "loss": 0.272,
+      "step": 1055
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.01266191523761e-05,
+      "loss": 0.2663,
+      "step": 1056
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.010677295667157e-05,
+      "loss": 0.2585,
+      "step": 1057
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.008691175379042e-05,
+      "loss": 0.2843,
+      "step": 1058
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.006703556346291e-05,
+      "loss": 0.2816,
+      "step": 1059
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.0047144405434175e-05,
+      "loss": 0.2867,
+      "step": 1060
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.0027238299464256e-05,
+      "loss": 0.266,
+      "step": 1061
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.0007317265328e-05,
+      "loss": 0.2579,
+      "step": 1062
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.998738132281512e-05,
+      "loss": 0.2592,
+      "step": 1063
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.9967430491730126e-05,
+      "loss": 0.2877,
+      "step": 1064
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.994746479189231e-05,
+      "loss": 0.306,
+      "step": 1065
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.9927484243135735e-05,
+      "loss": 0.2961,
+      "step": 1066
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.990748886530924e-05,
+      "loss": 0.2833,
+      "step": 1067
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.988747867827636e-05,
+      "loss": 0.2656,
+      "step": 1068
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.986745370191536e-05,
+      "loss": 0.2766,
+      "step": 1069
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.9847413956119206e-05,
+      "loss": 0.2618,
+      "step": 1070
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.9827359460795524e-05,
+      "loss": 0.272,
+      "step": 1071
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.980729023586658e-05,
+      "loss": 0.2795,
+      "step": 1072
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.978720630126929e-05,
+      "loss": 0.2819,
+      "step": 1073
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.976710767695519e-05,
+      "loss": 0.2851,
+      "step": 1074
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.974699438289039e-05,
+      "loss": 0.2735,
+      "step": 1075
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.972686643905558e-05,
+      "loss": 0.2901,
+      "step": 1076
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.9706723865446e-05,
+      "loss": 0.2652,
+      "step": 1077
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.968656668207144e-05,
+      "loss": 0.2518,
+      "step": 1078
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.966639490895617e-05,
+      "loss": 0.2513,
+      "step": 1079
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.9646208566138995e-05,
+      "loss": 0.27,
+      "step": 1080
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.962600767367317e-05,
+      "loss": 0.2729,
+      "step": 1081
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.9605792251626395e-05,
+      "loss": 0.271,
+      "step": 1082
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.958556232008082e-05,
+      "loss": 0.285,
+      "step": 1083
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.956531789913301e-05,
+      "loss": 0.2776,
+      "step": 1084
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.95450590088939e-05,
+      "loss": 0.2933,
+      "step": 1085
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.952478566948885e-05,
+      "loss": 0.2597,
+      "step": 1086
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.9504497901057514e-05,
+      "loss": 0.2511,
+      "step": 1087
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.9484195723753905e-05,
+      "loss": 0.2274,
+      "step": 1088
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.9463879157746364e-05,
+      "loss": 0.2985,
+      "step": 1089
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.9443548223217516e-05,
+      "loss": 0.2535,
+      "step": 1090
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.9423202940364246e-05,
+      "loss": 0.2498,
+      "step": 1091
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.940284332939771e-05,
+      "loss": 0.2683,
+      "step": 1092
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.938246941054329e-05,
+      "loss": 0.281,
+      "step": 1093
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.936208120404058e-05,
+      "loss": 0.314,
+      "step": 1094
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.9341678730143386e-05,
+      "loss": 0.2833,
+      "step": 1095
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.932126200911964e-05,
+      "loss": 0.265,
+      "step": 1096
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.93008310612515e-05,
+      "loss": 0.2896,
+      "step": 1097
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.9280385906835184e-05,
+      "loss": 0.2851,
+      "step": 1098
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.925992656618106e-05,
+      "loss": 0.2856,
+      "step": 1099
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.9239453059613604e-05,
+      "loss": 0.3038,
+      "step": 1100
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.921896540747133e-05,
+      "loss": 0.3109,
+      "step": 1101
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.9198463630106806e-05,
+      "loss": 0.2904,
+      "step": 1102
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.917794774788667e-05,
+      "loss": 0.2731,
+      "step": 1103
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.915741778119153e-05,
+      "loss": 0.3188,
+      "step": 1104
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.9136873750416006e-05,
+      "loss": 0.2877,
+      "step": 1105
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.9116315675968685e-05,
+      "loss": 0.271,
+      "step": 1106
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.909574357827211e-05,
+      "loss": 0.2947,
+      "step": 1107
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.9075157477762744e-05,
+      "loss": 0.2552,
+      "step": 1108
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.9054557394890974e-05,
+      "loss": 0.2647,
+      "step": 1109
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.903394335012106e-05,
+      "loss": 0.2785,
+      "step": 1110
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.901331536393115e-05,
+      "loss": 0.2778,
+      "step": 1111
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.8992673456813233e-05,
+      "loss": 0.2829,
+      "step": 1112
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.897201764927313e-05,
+      "loss": 0.2676,
+      "step": 1113
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.8951347961830465e-05,
+      "loss": 0.2832,
+      "step": 1114
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.8930664415018644e-05,
+      "loss": 0.2332,
+      "step": 1115
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.890996702938486e-05,
+      "loss": 0.2867,
+      "step": 1116
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.888925582549006e-05,
+      "loss": 0.2643,
+      "step": 1117
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.886853082390888e-05,
+      "loss": 0.2859,
+      "step": 1118
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.884779204522968e-05,
+      "loss": 0.2637,
+      "step": 1119
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.882703951005454e-05,
+      "loss": 0.2856,
+      "step": 1120
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.8806273238999156e-05,
+      "loss": 0.2713,
+      "step": 1121
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.87854932526929e-05,
+      "loss": 0.2657,
+      "step": 1122
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.876469957177874e-05,
+      "loss": 0.2811,
+      "step": 1123
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.874389221691329e-05,
+      "loss": 0.2986,
+      "step": 1124
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.872307120876672e-05,
+      "loss": 0.2683,
+      "step": 1125
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.870223656802277e-05,
+      "loss": 0.2634,
+      "step": 1126
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.868138831537872e-05,
+      "loss": 0.3144,
+      "step": 1127
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.866052647154537e-05,
+      "loss": 0.2547,
+      "step": 1128
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.863965105724703e-05,
+      "loss": 0.3147,
+      "step": 1129
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.861876209322148e-05,
+      "loss": 0.2728,
+      "step": 1130
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.8597859600219974e-05,
+      "loss": 0.2761,
+      "step": 1131
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.857694359900719e-05,
+      "loss": 0.2454,
+      "step": 1132
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.855601411036125e-05,
+      "loss": 0.266,
+      "step": 1133
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.853507115507364e-05,
+      "loss": 0.3202,
+      "step": 1134
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.851411475394925e-05,
+      "loss": 0.2638,
+      "step": 1135
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.849314492780632e-05,
+      "loss": 0.2882,
+      "step": 1136
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.847216169747642e-05,
+      "loss": 0.2953,
+      "step": 1137
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.845116508380445e-05,
+      "loss": 0.3025,
+      "step": 1138
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.843015510764859e-05,
+      "loss": 0.3034,
+      "step": 1139
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.84091317898803e-05,
+      "loss": 0.267,
+      "step": 1140
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.838809515138431e-05,
+      "loss": 0.2684,
+      "step": 1141
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.836704521305856e-05,
+      "loss": 0.2882,
+      "step": 1142
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.8345981995814186e-05,
+      "loss": 0.2733,
+      "step": 1143
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.832490552057557e-05,
+      "loss": 0.2933,
+      "step": 1144
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.830381580828022e-05,
+      "loss": 0.2689,
+      "step": 1145
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.828271287987881e-05,
+      "loss": 0.2434,
+      "step": 1146
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.826159675633514e-05,
+      "loss": 0.277,
+      "step": 1147
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.8240467458626115e-05,
+      "loss": 0.298,
+      "step": 1148
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.821932500774175e-05,
+      "loss": 0.2983,
+      "step": 1149
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.819816942468508e-05,
+      "loss": 0.3237,
+      "step": 1150
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.8177000730472226e-05,
+      "loss": 0.2948,
+      "step": 1151
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.815581894613231e-05,
+      "loss": 0.2796,
+      "step": 1152
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.81346240927075e-05,
+      "loss": 0.2885,
+      "step": 1153
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.8113416191252874e-05,
+      "loss": 0.3103,
+      "step": 1154
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.809219526283654e-05,
+      "loss": 0.2719,
+      "step": 1155
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.8070961328539525e-05,
+      "loss": 0.2716,
+      "step": 1156
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.804971440945575e-05,
+      "loss": 0.2643,
+      "step": 1157
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.8028454526692064e-05,
+      "loss": 0.2701,
+      "step": 1158
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.800718170136819e-05,
+      "loss": 0.2758,
+      "step": 1159
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.7985895954616714e-05,
+      "loss": 0.2901,
+      "step": 1160
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.796459730758304e-05,
+      "loss": 0.2836,
+      "step": 1161
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.7943285781425386e-05,
+      "loss": 0.2437,
+      "step": 1162
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.79219613973148e-05,
+      "loss": 0.2943,
+      "step": 1163
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.7900624176435066e-05,
+      "loss": 0.2858,
+      "step": 1164
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.787927413998272e-05,
+      "loss": 0.2514,
+      "step": 1165
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.785791130916706e-05,
+      "loss": 0.3013,
+      "step": 1166
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.7836535705210065e-05,
+      "loss": 0.2858,
+      "step": 1167
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.781514734934641e-05,
+      "loss": 0.317,
+      "step": 1168
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.779374626282346e-05,
+      "loss": 0.3049,
+      "step": 1169
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.7772332466901195e-05,
+      "loss": 0.2863,
+      "step": 1170
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.7750905982852226e-05,
+      "loss": 0.2707,
+      "step": 1171
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.772946683196179e-05,
+      "loss": 0.2904,
+      "step": 1172
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.770801503552768e-05,
+      "loss": 0.2793,
+      "step": 1173
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.768655061486028e-05,
+      "loss": 0.26,
+      "step": 1174
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.766507359128248e-05,
+      "loss": 0.2575,
+      "step": 1175
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.764358398612973e-05,
+      "loss": 0.2411,
+      "step": 1176
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.762208182074994e-05,
+      "loss": 0.263,
+      "step": 1177
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.760056711650353e-05,
+      "loss": 0.2839,
+      "step": 1178
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.7579039894763356e-05,
+      "loss": 0.2852,
+      "step": 1179
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.755750017691471e-05,
+      "loss": 0.2771,
+      "step": 1180
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.753594798435531e-05,
+      "loss": 0.2935,
+      "step": 1181
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.751438333849525e-05,
+      "loss": 0.269,
+      "step": 1182
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.749280626075701e-05,
+      "loss": 0.2577,
+      "step": 1183
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7471216772575416e-05,
+      "loss": 0.2609,
+      "step": 1184
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7449614895397613e-05,
+      "loss": 0.266,
+      "step": 1185
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7428000650683054e-05,
+      "loss": 0.2552,
+      "step": 1186
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.74063740599035e-05,
+      "loss": 0.249,
+      "step": 1187
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.738473514454297e-05,
+      "loss": 0.2934,
+      "step": 1188
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7363083926097694e-05,
+      "loss": 0.3069,
+      "step": 1189
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.734142042607616e-05,
+      "loss": 0.2822,
+      "step": 1190
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.731974466599904e-05,
+      "loss": 0.2723,
+      "step": 1191
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.72980566673992e-05,
+      "loss": 0.2974,
+      "step": 1192
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.727635645182164e-05,
+      "loss": 0.2301,
+      "step": 1193
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.725464404082351e-05,
+      "loss": 0.2795,
+      "step": 1194
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7232919455974084e-05,
+      "loss": 0.2946,
+      "step": 1195
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.721118271885472e-05,
+      "loss": 0.2495,
+      "step": 1196
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.7189433851058844e-05,
+      "loss": 0.2803,
+      "step": 1197
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.716767287419193e-05,
+      "loss": 0.3004,
+      "step": 1198
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.714589980987151e-05,
+      "loss": 0.3005,
+      "step": 1199
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.712411467972708e-05,
+      "loss": 0.2712,
+      "step": 1200
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.710231750540015e-05,
+      "loss": 0.3069,
+      "step": 1201
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.70805083085442e-05,
+      "loss": 0.2529,
+      "step": 1202
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.705868711082463e-05,
+      "loss": 0.274,
+      "step": 1203
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.7036853933918784e-05,
+      "loss": 0.2884,
+      "step": 1204
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.70150087995159e-05,
+      "loss": 0.2574,
+      "step": 1205
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.6993151729317075e-05,
+      "loss": 0.2762,
+      "step": 1206
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.6971282745035287e-05,
+      "loss": 0.2709,
+      "step": 1207
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.694940186839535e-05,
+      "loss": 0.2835,
+      "step": 1208
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.692750912113387e-05,
+      "loss": 0.2733,
+      "step": 1209
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.690560452499926e-05,
+      "loss": 0.2932,
+      "step": 1210
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.6883688101751724e-05,
+      "loss": 0.2656,
+      "step": 1211
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.686175987316317e-05,
+      "loss": 0.2797,
+      "step": 1212
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.683981986101727e-05,
+      "loss": 0.3308,
+      "step": 1213
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.681786808710938e-05,
+      "loss": 0.2778,
+      "step": 1214
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.6795904573246564e-05,
+      "loss": 0.2894,
+      "step": 1215
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.677392934124752e-05,
+      "loss": 0.2787,
+      "step": 1216
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.675194241294261e-05,
+      "loss": 0.2569,
+      "step": 1217
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.672994381017379e-05,
+      "loss": 0.2754,
+      "step": 1218
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.670793355479466e-05,
+      "loss": 0.2716,
+      "step": 1219
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.668591166867035e-05,
+      "loss": 0.256,
+      "step": 1220
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.666387817367755e-05,
+      "loss": 0.2741,
+      "step": 1221
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.6641833091704506e-05,
+      "loss": 0.262,
+      "step": 1222
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.661977644465094e-05,
+      "loss": 0.3074,
+      "step": 1223
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.659770825442812e-05,
+      "loss": 0.2847,
+      "step": 1224
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.657562854295871e-05,
+      "loss": 0.2885,
+      "step": 1225
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.655353733217686e-05,
+      "loss": 0.289,
+      "step": 1226
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.653143464402816e-05,
+      "loss": 0.2853,
+      "step": 1227
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.650932050046956e-05,
+      "loss": 0.3044,
+      "step": 1228
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.6487194923469406e-05,
+      "loss": 0.2703,
+      "step": 1229
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.64650579350074e-05,
+      "loss": 0.2806,
+      "step": 1230
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.644290955707461e-05,
+      "loss": 0.2852,
+      "step": 1231
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.6420749811673376e-05,
+      "loss": 0.2848,
+      "step": 1232
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.6398578720817354e-05,
+      "loss": 0.2705,
+      "step": 1233
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.6376396306531454e-05,
+      "loss": 0.2831,
+      "step": 1234
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.635420259085186e-05,
+      "loss": 0.3117,
+      "step": 1235
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.633199759582596e-05,
+      "loss": 0.3039,
+      "step": 1236
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.630978134351236e-05,
+      "loss": 0.2712,
+      "step": 1237
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.628755385598084e-05,
+      "loss": 0.2735,
+      "step": 1238
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.626531515531234e-05,
+      "loss": 0.265,
+      "step": 1239
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.624306526359896e-05,
+      "loss": 0.27,
+      "step": 1240
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.622080420294389e-05,
+      "loss": 0.2758,
+      "step": 1241
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.619853199546144e-05,
+      "loss": 0.278,
+      "step": 1242
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.617624866327696e-05,
+      "loss": 0.2783,
+      "step": 1243
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.6153954228526894e-05,
+      "loss": 0.2628,
+      "step": 1244
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.6131648713358676e-05,
+      "loss": 0.2951,
+      "step": 1245
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.610933213993077e-05,
+      "loss": 0.2586,
+      "step": 1246
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.608700453041263e-05,
+      "loss": 0.2926,
+      "step": 1247
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.6064665906984657e-05,
+      "loss": 0.2934,
+      "step": 1248
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.6042316291838184e-05,
+      "loss": 0.2876,
+      "step": 1249
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.601995570717551e-05,
+      "loss": 0.2748,
+      "step": 1250
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.5997584175209773e-05,
+      "loss": 0.2476,
+      "step": 1251
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.597520171816503e-05,
+      "loss": 0.2917,
+      "step": 1252
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.595280835827617e-05,
+      "loss": 0.2956,
+      "step": 1253
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.5930404117788905e-05,
+      "loss": 0.2611,
+      "step": 1254
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.5907989018959795e-05,
+      "loss": 0.2856,
+      "step": 1255
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.588556308405614e-05,
+      "loss": 0.2769,
+      "step": 1256
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.586312633535602e-05,
+      "loss": 0.282,
+      "step": 1257
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.584067879514827e-05,
+      "loss": 0.29,
+      "step": 1258
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.581822048573244e-05,
+      "loss": 0.2661,
+      "step": 1259
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.579575142941878e-05,
+      "loss": 0.2661,
+      "step": 1260
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.577327164852821e-05,
+      "loss": 0.27,
+      "step": 1261
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.5750781165392295e-05,
+      "loss": 0.3133,
+      "step": 1262
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.572828000235325e-05,
+      "loss": 0.2914,
+      "step": 1263
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.570576818176389e-05,
+      "loss": 0.2875,
+      "step": 1264
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.568324572598761e-05,
+      "loss": 0.2942,
+      "step": 1265
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.566071265739839e-05,
+      "loss": 0.2625,
+      "step": 1266
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.563816899838075e-05,
+      "loss": 0.2695,
+      "step": 1267
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.5615614771329706e-05,
+      "loss": 0.2741,
+      "step": 1268
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.5593049998650805e-05,
+      "loss": 0.264,
+      "step": 1269
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.557047470276004e-05,
+      "loss": 0.2519,
+      "step": 1270
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.5547888906083876e-05,
+      "loss": 0.2685,
+      "step": 1271
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.552529263105921e-05,
+      "loss": 0.312,
+      "step": 1272
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.550268590013333e-05,
+      "loss": 0.2566,
+      "step": 1273
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.5480068735763947e-05,
+      "loss": 0.2671,
+      "step": 1274
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.545744116041909e-05,
+      "loss": 0.2774,
+      "step": 1275
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.543480319657718e-05,
+      "loss": 0.2797,
+      "step": 1276
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.541215486672691e-05,
+      "loss": 0.2675,
+      "step": 1277
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.538949619336731e-05,
+      "loss": 0.2573,
+      "step": 1278
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.5366827199007656e-05,
+      "loss": 0.2714,
+      "step": 1279
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.5344147906167497e-05,
+      "loss": 0.2568,
+      "step": 1280
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.532145833737659e-05,
+      "loss": 0.2561,
+      "step": 1281
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.5298758515174946e-05,
+      "loss": 0.267,
+      "step": 1282
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.527604846211271e-05,
+      "loss": 0.2832,
+      "step": 1283
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.5253328200750224e-05,
+      "loss": 0.2559,
+      "step": 1284
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.523059775365793e-05,
+      "loss": 0.2829,
+      "step": 1285
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.520785714341645e-05,
+      "loss": 0.2662,
+      "step": 1286
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.518510639261646e-05,
+      "loss": 0.2714,
+      "step": 1287
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.516234552385871e-05,
+      "loss": 0.2808,
+      "step": 1288
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5139574559754005e-05,
+      "loss": 0.2846,
+      "step": 1289
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.511679352292322e-05,
+      "loss": 0.2633,
+      "step": 1290
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.509400243599716e-05,
+      "loss": 0.2912,
+      "step": 1291
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.507120132161667e-05,
+      "loss": 0.3025,
+      "step": 1292
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.504839020243254e-05,
+      "loss": 0.2723,
+      "step": 1293
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5025569101105506e-05,
+      "loss": 0.2531,
+      "step": 1294
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.50027380403062e-05,
+      "loss": 0.2927,
+      "step": 1295
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.497989704271517e-05,
+      "loss": 0.3019,
+      "step": 1296
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.495704613102283e-05,
+      "loss": 0.2917,
+      "step": 1297
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4934185327929436e-05,
+      "loss": 0.2484,
+      "step": 1298
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.491131465614507e-05,
+      "loss": 0.2993,
+      "step": 1299
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.488843413838963e-05,
+      "loss": 0.2588,
+      "step": 1300
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4865543797392776e-05,
+      "loss": 0.2536,
+      "step": 1301
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.484264365589394e-05,
+      "loss": 0.2783,
+      "step": 1302
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.48197337366423e-05,
+      "loss": 0.2823,
+      "step": 1303
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.479681406239671e-05,
+      "loss": 0.305,
+      "step": 1304
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.477388465592576e-05,
+      "loss": 0.2816,
+      "step": 1305
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.475094554000768e-05,
+      "loss": 0.2744,
+      "step": 1306
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4727996737430346e-05,
+      "loss": 0.2785,
+      "step": 1307
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.470503827099127e-05,
+      "loss": 0.2678,
+      "step": 1308
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4682070163497555e-05,
+      "loss": 0.2893,
+      "step": 1309
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4659092437765885e-05,
+      "loss": 0.274,
+      "step": 1310
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4636105116622504e-05,
+      "loss": 0.2502,
+      "step": 1311
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4613108222903155e-05,
+      "loss": 0.2595,
+      "step": 1312
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.459010177945316e-05,
+      "loss": 0.2665,
+      "step": 1313
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.456708580912725e-05,
+      "loss": 0.2951,
+      "step": 1314
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.454406033478966e-05,
+      "loss": 0.2685,
+      "step": 1315
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.452102537931408e-05,
+      "loss": 0.3223,
+      "step": 1316
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.449798096558358e-05,
+      "loss": 0.2773,
+      "step": 1317
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.447492711649067e-05,
+      "loss": 0.2928,
+      "step": 1318
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.4451863854937174e-05,
+      "loss": 0.2627,
+      "step": 1319
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.442879120383433e-05,
+      "loss": 0.258,
+      "step": 1320
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.4405709186102665e-05,
+      "loss": 0.2797,
+      "step": 1321
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.438261782467202e-05,
+      "loss": 0.2664,
+      "step": 1322
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.435951714248152e-05,
+      "loss": 0.2616,
+      "step": 1323
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.433640716247956e-05,
+      "loss": 0.2774,
+      "step": 1324
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.4313287907623735e-05,
+      "loss": 0.2808,
+      "step": 1325
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.4290159400880894e-05,
+      "loss": 0.2597,
+      "step": 1326
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.426702166522707e-05,
+      "loss": 0.2479,
+      "step": 1327
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.4243874723647435e-05,
+      "loss": 0.2604,
+      "step": 1328
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.422071859913634e-05,
+      "loss": 0.275,
+      "step": 1329
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.419755331469725e-05,
+      "loss": 0.2815,
+      "step": 1330
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.4174378893342714e-05,
+      "loss": 0.2526,
+      "step": 1331
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.4151195358094365e-05,
+      "loss": 0.2703,
+      "step": 1332
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.412800273198291e-05,
+      "loss": 0.2561,
+      "step": 1333
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.410480103804806e-05,
+      "loss": 0.3189,
+      "step": 1334
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.408159029933854e-05,
+      "loss": 0.2483,
+      "step": 1335
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.4058370538912065e-05,
+      "loss": 0.2878,
+      "step": 1336
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.4035141779835324e-05,
+      "loss": 0.2854,
+      "step": 1337
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.401190404518392e-05,
+      "loss": 0.2783,
+      "step": 1338
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3988657358042397e-05,
+      "loss": 0.291,
+      "step": 1339
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.396540174150416e-05,
+      "loss": 0.2931,
+      "step": 1340
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.394213721867153e-05,
+      "loss": 0.3038,
+      "step": 1341
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.391886381265564e-05,
+      "loss": 0.2763,
+      "step": 1342
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.389558154657645e-05,
+      "loss": 0.3144,
+      "step": 1343
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.387229044356274e-05,
+      "loss": 0.2923,
+      "step": 1344
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.384899052675206e-05,
+      "loss": 0.2859,
+      "step": 1345
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.382568181929072e-05,
+      "loss": 0.2866,
+      "step": 1346
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3802364344333735e-05,
+      "loss": 0.2768,
+      "step": 1347
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.377903812504487e-05,
+      "loss": 0.254,
+      "step": 1348
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3755703184596554e-05,
+      "loss": 0.2498,
+      "step": 1349
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.373235954616989e-05,
+      "loss": 0.2952,
+      "step": 1350
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3709007232954604e-05,
+      "loss": 0.2686,
+      "step": 1351
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3685646268149056e-05,
+      "loss": 0.2683,
+      "step": 1352
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.366227667496021e-05,
+      "loss": 0.291,
+      "step": 1353
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.363889847660356e-05,
+      "loss": 0.2743,
+      "step": 1354
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3615511696303204e-05,
+      "loss": 0.2584,
+      "step": 1355
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3592116357291714e-05,
+      "loss": 0.2694,
+      "step": 1356
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3568712482810205e-05,
+      "loss": 0.2582,
+      "step": 1357
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3545300096108247e-05,
+      "loss": 0.277,
+      "step": 1358
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.352187922044387e-05,
+      "loss": 0.305,
+      "step": 1359
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.349844987908355e-05,
+      "loss": 0.2781,
+      "step": 1360
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.347501209530215e-05,
+      "loss": 0.2732,
+      "step": 1361
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3451565892382936e-05,
+      "loss": 0.2758,
+      "step": 1362
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.342811129361754e-05,
+      "loss": 0.2813,
+      "step": 1363
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.340464832230592e-05,
+      "loss": 0.2986,
+      "step": 1364
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.338117700175637e-05,
+      "loss": 0.2765,
+      "step": 1365
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.335769735528546e-05,
+      "loss": 0.2613,
+      "step": 1366
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.333420940621805e-05,
+      "loss": 0.286,
+      "step": 1367
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.331071317788722e-05,
+      "loss": 0.2576,
+      "step": 1368
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.328720869363431e-05,
+      "loss": 0.2549,
+      "step": 1369
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.326369597680883e-05,
+      "loss": 0.2595,
+      "step": 1370
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3240175050768484e-05,
+      "loss": 0.245,
+      "step": 1371
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3216645938879135e-05,
+      "loss": 0.282,
+      "step": 1372
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.319310866451476e-05,
+      "loss": 0.2612,
+      "step": 1373
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3169563251057476e-05,
+      "loss": 0.2572,
+      "step": 1374
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3146009721897445e-05,
+      "loss": 0.2764,
+      "step": 1375
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.312244810043292e-05,
+      "loss": 0.2913,
+      "step": 1376
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.309887841007019e-05,
+      "loss": 0.2619,
+      "step": 1377
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.307530067422354e-05,
+      "loss": 0.2922,
+      "step": 1378
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.305171491631528e-05,
+      "loss": 0.261,
+      "step": 1379
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3028121159775656e-05,
+      "loss": 0.3187,
+      "step": 1380
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.300451942804289e-05,
+      "loss": 0.2784,
+      "step": 1381
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.29809097445631e-05,
+      "loss": 0.3012,
+      "step": 1382
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2957292132790306e-05,
+      "loss": 0.278,
+      "step": 1383
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2933666616186437e-05,
+      "loss": 0.2429,
+      "step": 1384
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.291003321822124e-05,
+      "loss": 0.2913,
+      "step": 1385
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.288639196237229e-05,
+      "loss": 0.2917,
+      "step": 1386
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2862742872125005e-05,
+      "loss": 0.286,
+      "step": 1387
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.283908597097254e-05,
+      "loss": 0.2767,
+      "step": 1388
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.281542128241583e-05,
+      "loss": 0.2747,
+      "step": 1389
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2791748829963564e-05,
+      "loss": 0.2718,
+      "step": 1390
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2768068637132106e-05,
+      "loss": 0.2619,
+      "step": 1391
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2744380727445546e-05,
+      "loss": 0.2662,
+      "step": 1392
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.272068512443561e-05,
+      "loss": 0.3032,
+      "step": 1393
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.26969818516417e-05,
+      "loss": 0.2579,
+      "step": 1394
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.267327093261079e-05,
+      "loss": 0.3057,
+      "step": 1395
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.2649552390897494e-05,
+      "loss": 0.2563,
+      "step": 1396
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.262582625006398e-05,
+      "loss": 0.2912,
+      "step": 1397
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.260209253367996e-05,
+      "loss": 0.2966,
+      "step": 1398
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.257835126532268e-05,
+      "loss": 0.2646,
+      "step": 1399
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.25546024685769e-05,
+      "loss": 0.2985,
+      "step": 1400
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.2530846167034826e-05,
+      "loss": 0.2691,
+      "step": 1401
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2507082384296155e-05,
+      "loss": 0.302,
+      "step": 1402
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.248331114396798e-05,
+      "loss": 0.2833,
+      "step": 1403
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.245953246966484e-05,
+      "loss": 0.2821,
+      "step": 1404
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2435746385008626e-05,
+      "loss": 0.2461,
+      "step": 1405
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.241195291362862e-05,
+      "loss": 0.2626,
+      "step": 1406
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.238815207916141e-05,
+      "loss": 0.2756,
+      "step": 1407
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.236434390525093e-05,
+      "loss": 0.2873,
+      "step": 1408
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.234052841554839e-05,
+      "loss": 0.2705,
+      "step": 1409
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.231670563371226e-05,
+      "loss": 0.3063,
+      "step": 1410
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.229287558340827e-05,
+      "loss": 0.3054,
+      "step": 1411
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.226903828830935e-05,
+      "loss": 0.2441,
+      "step": 1412
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.224519377209565e-05,
+      "loss": 0.2854,
+      "step": 1413
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.222134205845449e-05,
+      "loss": 0.2628,
+      "step": 1414
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2197483171080325e-05,
+      "loss": 0.3139,
+      "step": 1415
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.217361713367474e-05,
+      "loss": 0.2692,
+      "step": 1416
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.214974396994645e-05,
+      "loss": 0.2893,
+      "step": 1417
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.2125863703611196e-05,
+      "loss": 0.261,
+      "step": 1418
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.210197635839182e-05,
+      "loss": 0.266,
+      "step": 1419
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.207808195801819e-05,
+      "loss": 0.2756,
+      "step": 1420
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.205418052622716e-05,
+      "loss": 0.2702,
+      "step": 1421
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.20302720867626e-05,
+      "loss": 0.2613,
+      "step": 1422
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.200635666337531e-05,
+      "loss": 0.257,
+      "step": 1423
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.1982434279823054e-05,
+      "loss": 0.2824,
+      "step": 1424
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.1958504959870496e-05,
+      "loss": 0.2536,
+      "step": 1425
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.193456872728919e-05,
+      "loss": 0.3037,
+      "step": 1426
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.191062560585756e-05,
+      "loss": 0.279,
+      "step": 1427
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.1886675619360885e-05,
+      "loss": 0.2906,
+      "step": 1428
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.186271879159125e-05,
+      "loss": 0.2675,
+      "step": 1429
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.1838755146347535e-05,
+      "loss": 0.2795,
+      "step": 1430
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.1814784707435394e-05,
+      "loss": 0.2789,
+      "step": 1431
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.179080749866725e-05,
+      "loss": 0.2714,
+      "step": 1432
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.1766823543862215e-05,
+      "loss": 0.2903,
+      "step": 1433
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.174283286684613e-05,
+      "loss": 0.2665,
+      "step": 1434
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.171883549145149e-05,
+      "loss": 0.2967,
+      "step": 1435
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.1694831441517494e-05,
+      "loss": 0.2877,
+      "step": 1436
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.167082074088991e-05,
+      "loss": 0.273,
+      "step": 1437
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.1646803413421154e-05,
+      "loss": 0.2465,
+      "step": 1438
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.16227794829702e-05,
+      "loss": 0.2891,
+      "step": 1439
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.15987489734026e-05,
+      "loss": 0.2706,
+      "step": 1440
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.1574711908590434e-05,
+      "loss": 0.3024,
+      "step": 1441
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.1550668312412295e-05,
+      "loss": 0.2558,
+      "step": 1442
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.1526618208753255e-05,
+      "loss": 0.2392,
+      "step": 1443
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.1502561621504874e-05,
+      "loss": 0.2857,
+      "step": 1444
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.1478498574565137e-05,
+      "loss": 0.2699,
+      "step": 1445
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.145442909183843e-05,
+      "loss": 0.2969,
+      "step": 1446
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.143035319723556e-05,
+      "loss": 0.2671,
+      "step": 1447
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.140627091467371e-05,
+      "loss": 0.2992,
+      "step": 1448
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.138218226807636e-05,
+      "loss": 0.2783,
+      "step": 1449
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.135808728137338e-05,
+      "loss": 0.2893,
+      "step": 1450
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.133398597850087e-05,
+      "loss": 0.314,
+      "step": 1451
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.130987838340126e-05,
+      "loss": 0.2495,
+      "step": 1452
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.12857645200232e-05,
+      "loss": 0.2967,
+      "step": 1453
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.1261644412321564e-05,
+      "loss": 0.2609,
+      "step": 1454
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.1237518084257453e-05,
+      "loss": 0.2751,
+      "step": 1455
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.121338555979814e-05,
+      "loss": 0.2769,
+      "step": 1456
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.118924686291703e-05,
+      "loss": 0.2637,
+      "step": 1457
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.11651020175937e-05,
+      "loss": 0.2701,
+      "step": 1458
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.114095104781379e-05,
+      "loss": 0.2944,
+      "step": 1459
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.111679397756906e-05,
+      "loss": 0.2712,
+      "step": 1460
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.1092630830857305e-05,
+      "loss": 0.2854,
+      "step": 1461
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.1068461631682385e-05,
+      "loss": 0.2568,
+      "step": 1462
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.104428640405414e-05,
+      "loss": 0.3223,
+      "step": 1463
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.1020105171988426e-05,
+      "loss": 0.2488,
+      "step": 1464
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.0995917959507056e-05,
+      "loss": 0.2613,
+      "step": 1465
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.0971724790637764e-05,
+      "loss": 0.2805,
+      "step": 1466
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.0947525689414234e-05,
+      "loss": 0.261,
+      "step": 1467
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.0923320679876014e-05,
+      "loss": 0.2824,
+      "step": 1468
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.089910978606854e-05,
+      "loss": 0.2902,
+      "step": 1469
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.087489303204307e-05,
+      "loss": 0.2957,
+      "step": 1470
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.085067044185674e-05,
+      "loss": 0.2502,
+      "step": 1471
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.082644203957241e-05,
+      "loss": 0.2803,
+      "step": 1472
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.080220784925878e-05,
+      "loss": 0.2861,
+      "step": 1473
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.077796789499023e-05,
+      "loss": 0.2728,
+      "step": 1474
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.075372220084693e-05,
+      "loss": 0.2482,
+      "step": 1475
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.072947079091472e-05,
+      "loss": 0.276,
+      "step": 1476
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.070521368928513e-05,
+      "loss": 0.266,
+      "step": 1477
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.068095092005533e-05,
+      "loss": 0.2677,
+      "step": 1478
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.065668250732813e-05,
+      "loss": 0.2544,
+      "step": 1479
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.063240847521196e-05,
+      "loss": 0.3272,
+      "step": 1480
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.06081288478208e-05,
+      "loss": 0.2956,
+      "step": 1481
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.058384364927421e-05,
+      "loss": 0.2722,
+      "step": 1482
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.055955290369728e-05,
+      "loss": 0.3195,
+      "step": 1483
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.053525663522062e-05,
+      "loss": 0.2811,
+      "step": 1484
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0510954867980306e-05,
+      "loss": 0.2933,
+      "step": 1485
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0486647626117894e-05,
+      "loss": 0.2368,
+      "step": 1486
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0462334933780377e-05,
+      "loss": 0.2905,
+      "step": 1487
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0438016815120147e-05,
+      "loss": 0.2501,
+      "step": 1488
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0413693294295005e-05,
+      "loss": 0.2736,
+      "step": 1489
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.03893643954681e-05,
+      "loss": 0.2891,
+      "step": 1490
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.036503014280795e-05,
+      "loss": 0.2479,
+      "step": 1491
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.034069056048837e-05,
+      "loss": 0.2663,
+      "step": 1492
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0316345672688477e-05,
+      "loss": 0.2721,
+      "step": 1493
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.029199550359265e-05,
+      "loss": 0.3028,
+      "step": 1494
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0267640077390535e-05,
+      "loss": 0.2826,
+      "step": 1495
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0243279418276977e-05,
+      "loss": 0.2721,
+      "step": 1496
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0218913550452027e-05,
+      "loss": 0.2766,
+      "step": 1497
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0194542498120915e-05,
+      "loss": 0.2955,
+      "step": 1498
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.017016628549403e-05,
+      "loss": 0.2728,
+      "step": 1499
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0145784936786864e-05,
+      "loss": 0.2623,
+      "step": 1500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.012139847622003e-05,
+      "loss": 0.3031,
+      "step": 1501
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.009700692801921e-05,
+      "loss": 0.2746,
+      "step": 1502
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0072610316415146e-05,
+      "loss": 0.2547,
+      "step": 1503
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0048208665643603e-05,
+      "loss": 0.2773,
+      "step": 1504
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0023801999945357e-05,
+      "loss": 0.2862,
+      "step": 1505
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.9999390343566164e-05,
+      "loss": 0.2702,
+      "step": 1506
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.997497372075674e-05,
+      "loss": 0.2808,
+      "step": 1507
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.9950552155772743e-05,
+      "loss": 0.2662,
+      "step": 1508
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.9926125672874706e-05,
+      "loss": 0.2898,
+      "step": 1509
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.99016942963281e-05,
+      "loss": 0.2749,
+      "step": 1510
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.9877258050403212e-05,
+      "loss": 0.2713,
+      "step": 1511
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.9852816959375185e-05,
+      "loss": 0.2964,
+      "step": 1512
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.9828371047523984e-05,
+      "loss": 0.2878,
+      "step": 1513
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.9803920339134346e-05,
+      "loss": 0.2796,
+      "step": 1514
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.9779464858495786e-05,
+      "loss": 0.2591,
+      "step": 1515
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.9755004629902543e-05,
+      "loss": 0.2894,
+      "step": 1516
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.973053967765359e-05,
+      "loss": 0.274,
+      "step": 1517
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.970607002605259e-05,
+      "loss": 0.2817,
+      "step": 1518
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9681595699407864e-05,
+      "loss": 0.2823,
+      "step": 1519
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.965711672203239e-05,
+      "loss": 0.269,
+      "step": 1520
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9632633118243758e-05,
+      "loss": 0.2577,
+      "step": 1521
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.960814491236415e-05,
+      "loss": 0.2515,
+      "step": 1522
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.958365212872033e-05,
+      "loss": 0.2946,
+      "step": 1523
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.95591547916436e-05,
+      "loss": 0.2755,
+      "step": 1524
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9534652925469796e-05,
+      "loss": 0.2944,
+      "step": 1525
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9510146554539246e-05,
+      "loss": 0.2796,
+      "step": 1526
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9485635703196758e-05,
+      "loss": 0.2463,
+      "step": 1527
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.946112039579158e-05,
+      "loss": 0.305,
+      "step": 1528
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9436600656677405e-05,
+      "loss": 0.2859,
+      "step": 1529
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9412076510212305e-05,
+      "loss": 0.2815,
+      "step": 1530
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9387547980758756e-05,
+      "loss": 0.2537,
+      "step": 1531
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9363015092683566e-05,
+      "loss": 0.2489,
+      "step": 1532
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.933847787035787e-05,
+      "loss": 0.2557,
+      "step": 1533
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9313936338157154e-05,
+      "loss": 0.2769,
+      "step": 1534
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9289390520461123e-05,
+      "loss": 0.2853,
+      "step": 1535
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.926484044165379e-05,
+      "loss": 0.2819,
+      "step": 1536
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9240286126123355e-05,
+      "loss": 0.2721,
+      "step": 1537
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.921572759826227e-05,
+      "loss": 0.297,
+      "step": 1538
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.919116488246715e-05,
+      "loss": 0.2632,
+      "step": 1539
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9166598003138766e-05,
+      "loss": 0.2418,
+      "step": 1540
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.914202698468203e-05,
+      "loss": 0.3135,
+      "step": 1541
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9117451851505982e-05,
+      "loss": 0.2784,
+      "step": 1542
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9092872628023736e-05,
+      "loss": 0.2719,
+      "step": 1543
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.906828933865245e-05,
+      "loss": 0.2426,
+      "step": 1544
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9043702007813344e-05,
+      "loss": 0.2735,
+      "step": 1545
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.901911065993167e-05,
+      "loss": 0.2767,
+      "step": 1546
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.899451531943663e-05,
+      "loss": 0.2579,
+      "step": 1547
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.896991601076141e-05,
+      "loss": 0.3061,
+      "step": 1548
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.894531275834315e-05,
+      "loss": 0.2623,
+      "step": 1549
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.89207055866229e-05,
+      "loss": 0.3114,
+      "step": 1550
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.889609452004559e-05,
+      "loss": 0.2757,
+      "step": 1551
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.8871479583060034e-05,
+      "loss": 0.2974,
+      "step": 1552
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.8846860800118886e-05,
+      "loss": 0.2744,
+      "step": 1553
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.8822238195678625e-05,
+      "loss": 0.2769,
+      "step": 1554
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.8797611794199525e-05,
+      "loss": 0.2816,
+      "step": 1555
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.8772981620145623e-05,
+      "loss": 0.2422,
+      "step": 1556
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.8748347697984712e-05,
+      "loss": 0.2431,
+      "step": 1557
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.8723710052188314e-05,
+      "loss": 0.3001,
+      "step": 1558
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.8699068707231637e-05,
+      "loss": 0.2562,
+      "step": 1559
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.8674423687593565e-05,
+      "loss": 0.2834,
+      "step": 1560
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.8649775017756643e-05,
+      "loss": 0.2956,
+      "step": 1561
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.862512272220704e-05,
+      "loss": 0.2794,
+      "step": 1562
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.8600466825434517e-05,
+      "loss": 0.2685,
+      "step": 1563
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.8575807351932426e-05,
+      "loss": 0.2473,
+      "step": 1564
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.8551144326197655e-05,
+      "loss": 0.2749,
+      "step": 1565
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.8526477772730644e-05,
+      "loss": 0.2653,
+      "step": 1566
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.8501807716035305e-05,
+      "loss": 0.2836,
+      "step": 1567
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.8477134180619068e-05,
+      "loss": 0.2586,
+      "step": 1568
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.845245719099279e-05,
+      "loss": 0.2822,
+      "step": 1569
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.8427776771670766e-05,
+      "loss": 0.2511,
+      "step": 1570
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.8403092947170718e-05,
+      "loss": 0.2541,
+      "step": 1571
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.8378405742013713e-05,
+      "loss": 0.2673,
+      "step": 1572
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.8353715180724217e-05,
+      "loss": 0.2805,
+      "step": 1573
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8329021287830005e-05,
+      "loss": 0.295,
+      "step": 1574
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8304324087862157e-05,
+      "loss": 0.2686,
+      "step": 1575
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.827962360535506e-05,
+      "loss": 0.2781,
+      "step": 1576
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.825491986484634e-05,
+      "loss": 0.2877,
+      "step": 1577
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.823021289087689e-05,
+      "loss": 0.2841,
+      "step": 1578
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8205502707990773e-05,
+      "loss": 0.2284,
+      "step": 1579
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.818078934073527e-05,
+      "loss": 0.2856,
+      "step": 1580
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8156072813660818e-05,
+      "loss": 0.2779,
+      "step": 1581
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8131353151320994e-05,
+      "loss": 0.2494,
+      "step": 1582
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.810663037827249e-05,
+      "loss": 0.2651,
+      "step": 1583
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8081904519075082e-05,
+      "loss": 0.2729,
+      "step": 1584
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8057175598291613e-05,
+      "loss": 0.2716,
+      "step": 1585
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8032443640487975e-05,
+      "loss": 0.2881,
+      "step": 1586
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8007708670233073e-05,
+      "loss": 0.2576,
+      "step": 1587
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.7982970712098794e-05,
+      "loss": 0.2431,
+      "step": 1588
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.7958229790660017e-05,
+      "loss": 0.2723,
+      "step": 1589
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.7933485930494548e-05,
+      "loss": 0.2827,
+      "step": 1590
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.7908739156183107e-05,
+      "loss": 0.2509,
+      "step": 1591
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.7883989492309324e-05,
+      "loss": 0.2592,
+      "step": 1592
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.785923696345969e-05,
+      "loss": 0.2914,
+      "step": 1593
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.783448159422355e-05,
+      "loss": 0.2888,
+      "step": 1594
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.7809723409193057e-05,
+      "loss": 0.2967,
+      "step": 1595
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.778496243296317e-05,
+      "loss": 0.2515,
+      "step": 1596
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.7760198690131624e-05,
+      "loss": 0.2664,
+      "step": 1597
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.7735432205298906e-05,
+      "loss": 0.2573,
+      "step": 1598
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.7710663003068212e-05,
+      "loss": 0.2599,
+      "step": 1599
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.7685891108045436e-05,
+      "loss": 0.2725,
+      "step": 1600
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.7661116544839173e-05,
+      "loss": 0.2712,
+      "step": 1601
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.763633933806064e-05,
+      "loss": 0.2567,
+      "step": 1602
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.7611559512323704e-05,
+      "loss": 0.2908,
+      "step": 1603
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.7586777092244804e-05,
+      "loss": 0.2711,
+      "step": 1604
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.756199210244299e-05,
+      "loss": 0.2669,
+      "step": 1605
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.753720456753985e-05,
+      "loss": 0.2674,
+      "step": 1606
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.7512414512159484e-05,
+      "loss": 0.2403,
+      "step": 1607
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.7487621960928513e-05,
+      "loss": 0.2823,
+      "step": 1608
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.746282693847605e-05,
+      "loss": 0.2891,
+      "step": 1609
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.7438029469433635e-05,
+      "loss": 0.2705,
+      "step": 1610
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.7413229578435255e-05,
+      "loss": 0.2512,
+      "step": 1611
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.7388427290117296e-05,
+      "loss": 0.2347,
+      "step": 1612
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.736362262911854e-05,
+      "loss": 0.2734,
+      "step": 1613
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.7338815620080116e-05,
+      "loss": 0.2625,
+      "step": 1614
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.7314006287645466e-05,
+      "loss": 0.2949,
+      "step": 1615
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.728919465646037e-05,
+      "loss": 0.2936,
+      "step": 1616
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.7264380751172885e-05,
+      "loss": 0.2758,
+      "step": 1617
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.7239564596433314e-05,
+      "loss": 0.2822,
+      "step": 1618
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.7214746216894216e-05,
+      "loss": 0.2828,
+      "step": 1619
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.7189925637210323e-05,
+      "loss": 0.2928,
+      "step": 1620
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.7165102882038607e-05,
+      "loss": 0.2702,
+      "step": 1621
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.7140277976038154e-05,
+      "loss": 0.2759,
+      "step": 1622
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.71154509438702e-05,
+      "loss": 0.2741,
+      "step": 1623
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.7090621810198102e-05,
+      "loss": 0.267,
+      "step": 1624
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.7065790599687306e-05,
+      "loss": 0.2757,
+      "step": 1625
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.7040957337005305e-05,
+      "loss": 0.2738,
+      "step": 1626
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.701612204682165e-05,
+      "loss": 0.3378,
+      "step": 1627
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.6991284753807888e-05,
+      "loss": 0.286,
+      "step": 1628
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.696644548263757e-05,
+      "loss": 0.2921,
+      "step": 1629
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.6941604257986196e-05,
+      "loss": 0.2769,
+      "step": 1630
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.691676110453123e-05,
+      "loss": 0.2806,
+      "step": 1631
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.6891916046952025e-05,
+      "loss": 0.2566,
+      "step": 1632
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.6867069109929854e-05,
+      "loss": 0.2421,
+      "step": 1633
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.684222031814783e-05,
+      "loss": 0.2798,
+      "step": 1634
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.6817369696290932e-05,
+      "loss": 0.2974,
+      "step": 1635
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.6792517269045943e-05,
+      "loss": 0.2582,
+      "step": 1636
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.676766306110144e-05,
+      "loss": 0.2563,
+      "step": 1637
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.6742807097147764e-05,
+      "loss": 0.2355,
+      "step": 1638
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.671794940187702e-05,
+      "loss": 0.2711,
+      "step": 1639
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.669308999998301e-05,
+      "loss": 0.2325,
+      "step": 1640
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.6668228916161257e-05,
+      "loss": 0.2616,
+      "step": 1641
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.6643366175108937e-05,
+      "loss": 0.2612,
+      "step": 1642
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.6618501801524854e-05,
+      "loss": 0.2549,
+      "step": 1643
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 0.640147864818573,
+      "eval_runtime": 142.9689,
+      "eval_samples_per_second": 5.33,
+      "eval_steps_per_second": 0.336,
+      "step": 1643
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.659363582010948e-05,
+      "loss": 0.2112,
+      "step": 1644
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.6568768255564848e-05,
+      "loss": 0.1769,
+      "step": 1645
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.654389913259458e-05,
+      "loss": 0.1479,
+      "step": 1646
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.6519028475903833e-05,
+      "loss": 0.1553,
+      "step": 1647
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.6494156310199302e-05,
+      "loss": 0.1441,
+      "step": 1648
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.6469282660189183e-05,
+      "loss": 0.1507,
+      "step": 1649
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.644440755058312e-05,
+      "loss": 0.1497,
+      "step": 1650
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.6419531006092235e-05,
+      "loss": 0.1302,
+      "step": 1651
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.6394653051429064e-05,
+      "loss": 0.1278,
+      "step": 1652
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.6369773711307555e-05,
+      "loss": 0.1229,
+      "step": 1653
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.634489301044301e-05,
+      "loss": 0.1373,
+      "step": 1654
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.63200109735521e-05,
+      "loss": 0.1475,
+      "step": 1655
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.6295127625352823e-05,
+      "loss": 0.1479,
+      "step": 1656
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.627024299056447e-05,
+      "loss": 0.1396,
+      "step": 1657
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.624535709390762e-05,
+      "loss": 0.1248,
+      "step": 1658
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.622046996010409e-05,
+      "loss": 0.1265,
+      "step": 1659
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.6195581613876956e-05,
+      "loss": 0.1358,
+      "step": 1660
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.6170692079950466e-05,
+      "loss": 0.1346,
+      "step": 1661
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.6145801383050073e-05,
+      "loss": 0.1492,
+      "step": 1662
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.612090954790236e-05,
+      "loss": 0.1453,
+      "step": 1663
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.6096016599235064e-05,
+      "loss": 0.1367,
+      "step": 1664
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.6071122561777013e-05,
+      "loss": 0.1282,
+      "step": 1665
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.6046227460258126e-05,
+      "loss": 0.1335,
+      "step": 1666
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.602133131940937e-05,
+      "loss": 0.1209,
+      "step": 1667
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.5996434163962763e-05,
+      "loss": 0.1291,
+      "step": 1668
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.5971536018651305e-05,
+      "loss": 0.1292,
+      "step": 1669
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.5946636908209006e-05,
+      "loss": 0.1423,
+      "step": 1670
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.59217368573708e-05,
+      "loss": 0.1299,
+      "step": 1671
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.589683589087259e-05,
+      "loss": 0.139,
+      "step": 1672
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.5871934033451173e-05,
+      "loss": 0.1245,
+      "step": 1673
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.5847031309844232e-05,
+      "loss": 0.109,
+      "step": 1674
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.582212774479031e-05,
+      "loss": 0.1425,
+      "step": 1675
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.5797223363028794e-05,
+      "loss": 0.1375,
+      "step": 1676
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.577231818929987e-05,
+      "loss": 0.1304,
+      "step": 1677
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.5747412248344517e-05,
+      "loss": 0.1123,
+      "step": 1678
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.572250556490447e-05,
+      "loss": 0.1259,
+      "step": 1679
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.5697598163722213e-05,
+      "loss": 0.1333,
+      "step": 1680
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.567269006954094e-05,
+      "loss": 0.1389,
+      "step": 1681
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.5647781307104517e-05,
+      "loss": 0.1287,
+      "step": 1682
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.56228719011575e-05,
+      "loss": 0.124,
+      "step": 1683
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.5597961876445077e-05,
+      "loss": 0.1309,
+      "step": 1684
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.5573051257713034e-05,
+      "loss": 0.1216,
+      "step": 1685
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.5548140069707755e-05,
+      "loss": 0.139,
+      "step": 1686
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.5523228337176196e-05,
+      "loss": 0.1415,
+      "step": 1687
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.5498316084865857e-05,
+      "loss": 0.1436,
+      "step": 1688
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.5473403337524748e-05,
+      "loss": 0.1275,
+      "step": 1689
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.5448490119901364e-05,
+      "loss": 0.1324,
+      "step": 1690
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.542357645674468e-05,
+      "loss": 0.1273,
+      "step": 1691
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.5398662372804104e-05,
+      "loss": 0.1363,
+      "step": 1692
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.537374789282947e-05,
+      "loss": 0.1306,
+      "step": 1693
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.5348833041570995e-05,
+      "loss": 0.1263,
+      "step": 1694
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.532391784377927e-05,
+      "loss": 0.1246,
+      "step": 1695
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.529900232420524e-05,
+      "loss": 0.1379,
+      "step": 1696
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.5274086507600154e-05,
+      "loss": 0.1366,
+      "step": 1697
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.524917041871557e-05,
+      "loss": 0.1327,
+      "step": 1698
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.52242540823033e-05,
+      "loss": 0.124,
+      "step": 1699
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.5199337523115418e-05,
+      "loss": 0.1266,
+      "step": 1700
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.517442076590421e-05,
+      "loss": 0.1285,
+      "step": 1701
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.5149503835422167e-05,
+      "loss": 0.1305,
+      "step": 1702
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.512458675642193e-05,
+      "loss": 0.1253,
+      "step": 1703
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.5099669553656318e-05,
+      "loss": 0.1268,
+      "step": 1704
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.507475225187827e-05,
+      "loss": 0.1259,
+      "step": 1705
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.5049834875840783e-05,
+      "loss": 0.1278,
+      "step": 1706
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.502491745029698e-05,
+      "loss": 0.1327,
+      "step": 1707
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.5e-05,
+      "loss": 0.1247,
+      "step": 1708
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.4975082549703026e-05,
+      "loss": 0.123,
+      "step": 1709
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.4950165124159222e-05,
+      "loss": 0.1238,
+      "step": 1710
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.4925247748121737e-05,
+      "loss": 0.1252,
+      "step": 1711
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.4900330446343678e-05,
+      "loss": 0.1369,
+      "step": 1712
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.487541324357807e-05,
+      "loss": 0.1397,
+      "step": 1713
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.4850496164577842e-05,
+      "loss": 0.1416,
+      "step": 1714
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.4825579234095787e-05,
+      "loss": 0.1289,
+      "step": 1715
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.480066247688459e-05,
+      "loss": 0.1295,
+      "step": 1716
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.4775745917696712e-05,
+      "loss": 0.1382,
+      "step": 1717
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.4750829581284437e-05,
+      "loss": 0.1311,
+      "step": 1718
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.4725913492399848e-05,
+      "loss": 0.1247,
+      "step": 1719
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.4700997675794766e-05,
+      "loss": 0.1252,
+      "step": 1720
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.4676082156220736e-05,
+      "loss": 0.1284,
+      "step": 1721
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.465116695842901e-05,
+      "loss": 0.1221,
+      "step": 1722
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.4626252107170536e-05,
+      "loss": 0.1366,
+      "step": 1723
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.46013376271959e-05,
+      "loss": 0.1277,
+      "step": 1724
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.4576423543255325e-05,
+      "loss": 0.1255,
+      "step": 1725
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.455150988009864e-05,
+      "loss": 0.1451,
+      "step": 1726
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.452659666247525e-05,
+      "loss": 0.1275,
+      "step": 1727
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.4501683915134142e-05,
+      "loss": 0.1332,
+      "step": 1728
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.4476771662823803e-05,
+      "loss": 0.1453,
+      "step": 1729
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.4451859930292257e-05,
+      "loss": 0.1328,
+      "step": 1730
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.442694874228698e-05,
+      "loss": 0.1495,
+      "step": 1731
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.4402038123554933e-05,
+      "loss": 0.1364,
+      "step": 1732
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.4377128098842505e-05,
+      "loss": 0.1206,
+      "step": 1733
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.4352218692895485e-05,
+      "loss": 0.1303,
+      "step": 1734
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.432730993045907e-05,
+      "loss": 0.1329,
+      "step": 1735
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.4302401836277793e-05,
+      "loss": 0.1281,
+      "step": 1736
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.427749443509554e-05,
+      "loss": 0.123,
+      "step": 1737
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.4252587751655486e-05,
+      "loss": 0.1334,
+      "step": 1738
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.4227681810700133e-05,
+      "loss": 0.1274,
+      "step": 1739
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.4202776636971205e-05,
+      "loss": 0.1215,
+      "step": 1740
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.4177872255209686e-05,
+      "loss": 0.1408,
+      "step": 1741
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.415296869015577e-05,
+      "loss": 0.1427,
+      "step": 1742
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.4128065966548823e-05,
+      "loss": 0.1434,
+      "step": 1743
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.410316410912742e-05,
+      "loss": 0.1216,
+      "step": 1744
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.4078263142629206e-05,
+      "loss": 0.1373,
+      "step": 1745
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.4053363091791007e-05,
+      "loss": 0.1408,
+      "step": 1746
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.40284639813487e-05,
+      "loss": 0.1348,
+      "step": 1747
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.4003565836037246e-05,
+      "loss": 0.1372,
+      "step": 1748
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3978668680590635e-05,
+      "loss": 0.1416,
+      "step": 1749
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3953772539741877e-05,
+      "loss": 0.1224,
+      "step": 1750
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3928877438222993e-05,
+      "loss": 0.1407,
+      "step": 1751
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3903983400764942e-05,
+      "loss": 0.1237,
+      "step": 1752
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3879090452097648e-05,
+      "loss": 0.1341,
+      "step": 1753
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.385419861694993e-05,
+      "loss": 0.1263,
+      "step": 1754
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.382930792004953e-05,
+      "loss": 0.1337,
+      "step": 1755
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3804418386123047e-05,
+      "loss": 0.1376,
+      "step": 1756
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.377953003989591e-05,
+      "loss": 0.1302,
+      "step": 1757
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3754642906092393e-05,
+      "loss": 0.1305,
+      "step": 1758
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.372975700943554e-05,
+      "loss": 0.1295,
+      "step": 1759
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.370487237464719e-05,
+      "loss": 0.1281,
+      "step": 1760
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3679989026447906e-05,
+      "loss": 0.1387,
+      "step": 1761
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3655106989556998e-05,
+      "loss": 0.1283,
+      "step": 1762
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3630226288692455e-05,
+      "loss": 0.1343,
+      "step": 1763
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.360534694857094e-05,
+      "loss": 0.128,
+      "step": 1764
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3580468993907774e-05,
+      "loss": 0.1485,
+      "step": 1765
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3555592449416884e-05,
+      "loss": 0.1432,
+      "step": 1766
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3530717339810826e-05,
+      "loss": 0.1397,
+      "step": 1767
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.35058436898007e-05,
+      "loss": 0.1245,
+      "step": 1768
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.348097152409617e-05,
+      "loss": 0.1226,
+      "step": 1769
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3456100867405422e-05,
+      "loss": 0.1317,
+      "step": 1770
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.343123174443515e-05,
+      "loss": 0.1107,
+      "step": 1771
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3406364179890535e-05,
+      "loss": 0.1361,
+      "step": 1772
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3381498198475152e-05,
+      "loss": 0.129,
+      "step": 1773
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.335663382489108e-05,
+      "loss": 0.1318,
+      "step": 1774
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.333177108383875e-05,
+      "loss": 0.179,
+      "step": 1775
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3306910000016994e-05,
+      "loss": 0.1192,
+      "step": 1776
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3282050598122983e-05,
+      "loss": 0.1227,
+      "step": 1777
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.325719290285224e-05,
+      "loss": 0.1361,
+      "step": 1778
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.323233693889857e-05,
+      "loss": 0.1423,
+      "step": 1779
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3207482730954063e-05,
+      "loss": 0.154,
+      "step": 1780
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.318263030370907e-05,
+      "loss": 0.1393,
+      "step": 1781
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3157779681852164e-05,
+      "loss": 0.1436,
+      "step": 1782
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3132930890070145e-05,
+      "loss": 0.1355,
+      "step": 1783
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3108083953047974e-05,
+      "loss": 0.1389,
+      "step": 1784
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3083238895468775e-05,
+      "loss": 0.1288,
+      "step": 1785
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3058395742013814e-05,
+      "loss": 0.1341,
+      "step": 1786
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3033554517362442e-05,
+      "loss": 0.1253,
+      "step": 1787
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.300871524619212e-05,
+      "loss": 0.1384,
+      "step": 1788
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.2983877953178355e-05,
+      "loss": 0.1471,
+      "step": 1789
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.2959042662994697e-05,
+      "loss": 0.1273,
+      "step": 1790
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.29342094003127e-05,
+      "loss": 0.1242,
+      "step": 1791
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.2909378189801904e-05,
+      "loss": 0.1159,
+      "step": 1792
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.2884549056129802e-05,
+      "loss": 0.1466,
+      "step": 1793
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.285972202396185e-05,
+      "loss": 0.1116,
+      "step": 1794
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.2834897117961395e-05,
+      "loss": 0.1257,
+      "step": 1795
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2810074362789676e-05,
+      "loss": 0.1351,
+      "step": 1796
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2785253783105794e-05,
+      "loss": 0.1387,
+      "step": 1797
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.276043540356668e-05,
+      "loss": 0.1316,
+      "step": 1798
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2735619248827114e-05,
+      "loss": 0.1342,
+      "step": 1799
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2710805343539636e-05,
+      "loss": 0.1204,
+      "step": 1800
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2685993712354543e-05,
+      "loss": 0.1232,
+      "step": 1801
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2661184379919893e-05,
+      "loss": 0.1421,
+      "step": 1802
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2636377370881466e-05,
+      "loss": 0.1618,
+      "step": 1803
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2611572709882706e-05,
+      "loss": 0.1387,
+      "step": 1804
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2586770421564747e-05,
+      "loss": 0.1283,
+      "step": 1805
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.256197053056637e-05,
+      "loss": 0.1341,
+      "step": 1806
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2537173061523958e-05,
+      "loss": 0.142,
+      "step": 1807
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.251237803907149e-05,
+      "loss": 0.1414,
+      "step": 1808
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2487585487840522e-05,
+      "loss": 0.1227,
+      "step": 1809
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2462795432460156e-05,
+      "loss": 0.1326,
+      "step": 1810
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.243800789755701e-05,
+      "loss": 0.1317,
+      "step": 1811
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.2413222907755195e-05,
+      "loss": 0.1163,
+      "step": 1812
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.2388440487676302e-05,
+      "loss": 0.1371,
+      "step": 1813
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.2363660661939367e-05,
+      "loss": 0.135,
+      "step": 1814
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.233888345516084e-05,
+      "loss": 0.1349,
+      "step": 1815
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.231410889195457e-05,
+      "loss": 0.1297,
+      "step": 1816
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.22893369969318e-05,
+      "loss": 0.1375,
+      "step": 1817
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.2264567794701104e-05,
+      "loss": 0.1288,
+      "step": 1818
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.223980130986838e-05,
+      "loss": 0.1449,
+      "step": 1819
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.221503756703684e-05,
+      "loss": 0.129,
+      "step": 1820
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.219027659080695e-05,
+      "loss": 0.1343,
+      "step": 1821
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.216551840577646e-05,
+      "loss": 0.1245,
+      "step": 1822
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2140763036540312e-05,
+      "loss": 0.1448,
+      "step": 1823
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.211601050769068e-05,
+      "loss": 0.1277,
+      "step": 1824
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2091260843816892e-05,
+      "loss": 0.1355,
+      "step": 1825
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2066514069505455e-05,
+      "loss": 0.1235,
+      "step": 1826
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2041770209339982e-05,
+      "loss": 0.1455,
+      "step": 1827
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2017029287901212e-05,
+      "loss": 0.1502,
+      "step": 1828
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.199229132976694e-05,
+      "loss": 0.1307,
+      "step": 1829
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.1967556359512034e-05,
+      "loss": 0.133,
+      "step": 1830
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.19428244017084e-05,
+      "loss": 0.1412,
+      "step": 1831
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.1918095480924924e-05,
+      "loss": 0.1358,
+      "step": 1832
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.1893369621727518e-05,
+      "loss": 0.1265,
+      "step": 1833
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.186864684867901e-05,
+      "loss": 0.1328,
+      "step": 1834
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.1843927186339185e-05,
+      "loss": 0.1268,
+      "step": 1835
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.1819210659264737e-05,
+      "loss": 0.1273,
+      "step": 1836
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.179449729200923e-05,
+      "loss": 0.1495,
+      "step": 1837
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.1769787109123116e-05,
+      "loss": 0.1432,
+      "step": 1838
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.174508013515366e-05,
+      "loss": 0.1365,
+      "step": 1839
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.1720376394644946e-05,
+      "loss": 0.1311,
+      "step": 1840
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.1695675912137842e-05,
+      "loss": 0.1274,
+      "step": 1841
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.1670978712170008e-05,
+      "loss": 0.142,
+      "step": 1842
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.1646284819275792e-05,
+      "loss": 0.1316,
+      "step": 1843
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.162159425798629e-05,
+      "loss": 0.1375,
+      "step": 1844
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.159690705282929e-05,
+      "loss": 0.1516,
+      "step": 1845
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.157222322832924e-05,
+      "loss": 0.1279,
+      "step": 1846
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.154754280900722e-05,
+      "loss": 0.1415,
+      "step": 1847
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.1522865819380938e-05,
+      "loss": 0.1366,
+      "step": 1848
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.14981922839647e-05,
+      "loss": 0.1324,
+      "step": 1849
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.1473522227269366e-05,
+      "loss": 0.1401,
+      "step": 1850
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.1448855673802347e-05,
+      "loss": 0.1377,
+      "step": 1851
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.142419264806758e-05,
+      "loss": 0.1445,
+      "step": 1852
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.1399533174565485e-05,
+      "loss": 0.1326,
+      "step": 1853
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.1374877277792964e-05,
+      "loss": 0.125,
+      "step": 1854
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.135022498224336e-05,
+      "loss": 0.1337,
+      "step": 1855
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.1325576312406445e-05,
+      "loss": 0.1268,
+      "step": 1856
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.130093129276838e-05,
+      "loss": 0.1347,
+      "step": 1857
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.12762899478117e-05,
+      "loss": 0.1365,
+      "step": 1858
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.1251652302015297e-05,
+      "loss": 0.1469,
+      "step": 1859
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.1227018379854383e-05,
+      "loss": 0.1184,
+      "step": 1860
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.120238820580048e-05,
+      "loss": 0.1309,
+      "step": 1861
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.1177761804321378e-05,
+      "loss": 0.1133,
+      "step": 1862
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.115313919988112e-05,
+      "loss": 0.1746,
+      "step": 1863
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.112852041693997e-05,
+      "loss": 0.1272,
+      "step": 1864
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.1103905479954414e-05,
+      "loss": 0.1273,
+      "step": 1865
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.1079294413377103e-05,
+      "loss": 0.1544,
+      "step": 1866
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.105468724165685e-05,
+      "loss": 0.1385,
+      "step": 1867
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.1030083989238592e-05,
+      "loss": 0.1414,
+      "step": 1868
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.1005484680563374e-05,
+      "loss": 0.1291,
+      "step": 1869
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.0980889340068344e-05,
+      "loss": 0.1345,
+      "step": 1870
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.095629799218666e-05,
+      "loss": 0.1359,
+      "step": 1871
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.093171066134756e-05,
+      "loss": 0.1344,
+      "step": 1872
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.0907127371976276e-05,
+      "loss": 0.1336,
+      "step": 1873
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.088254814849402e-05,
+      "loss": 0.1359,
+      "step": 1874
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.0857973015317973e-05,
+      "loss": 0.147,
+      "step": 1875
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.083340199686124e-05,
+      "loss": 0.1266,
+      "step": 1876
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.0808835117532854e-05,
+      "loss": 0.1231,
+      "step": 1877
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.0784272401737732e-05,
+      "loss": 0.1288,
+      "step": 1878
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.0759713873876647e-05,
+      "loss": 0.1317,
+      "step": 1879
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.0735159558346215e-05,
+      "loss": 0.1233,
+      "step": 1880
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.0710609479538873e-05,
+      "loss": 0.1337,
+      "step": 1881
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.0686063661842848e-05,
+      "loss": 0.1411,
+      "step": 1882
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.0661522129642125e-05,
+      "loss": 0.1325,
+      "step": 1883
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.0636984907316447e-05,
+      "loss": 0.1278,
+      "step": 1884
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.0612452019241256e-05,
+      "loss": 0.1306,
+      "step": 1885
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.0587923489787704e-05,
+      "loss": 0.122,
+      "step": 1886
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.05633993433226e-05,
+      "loss": 0.1359,
+      "step": 1887
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.0538879604208422e-05,
+      "loss": 0.1271,
+      "step": 1888
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.0514364296803248e-05,
+      "loss": 0.1167,
+      "step": 1889
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.048985344546076e-05,
+      "loss": 0.1321,
+      "step": 1890
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.0465347074530213e-05,
+      "loss": 0.1104,
+      "step": 1891
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.0440845208356402e-05,
+      "loss": 0.1364,
+      "step": 1892
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.0416347871279677e-05,
+      "loss": 0.1308,
+      "step": 1893
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.0391855087635856e-05,
+      "loss": 0.1201,
+      "step": 1894
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.036736688175625e-05,
+      "loss": 0.1269,
+      "step": 1895
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.034288327796761e-05,
+      "loss": 0.1375,
+      "step": 1896
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.0318404300592135e-05,
+      "loss": 0.1262,
+      "step": 1897
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.029392997394742e-05,
+      "loss": 0.1242,
+      "step": 1898
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.0269460322346415e-05,
+      "loss": 0.1326,
+      "step": 1899
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.0244995370097463e-05,
+      "loss": 0.1249,
+      "step": 1900
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.0220535141504226e-05,
+      "loss": 0.1239,
+      "step": 1901
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.0196079660865663e-05,
+      "loss": 0.1252,
+      "step": 1902
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.0171628952476018e-05,
+      "loss": 0.1346,
+      "step": 1903
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.0147183040624818e-05,
+      "loss": 0.1214,
+      "step": 1904
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.0122741949596797e-05,
+      "loss": 0.1345,
+      "step": 1905
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.009830570367191e-05,
+      "loss": 0.1443,
+      "step": 1906
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.0073874327125296e-05,
+      "loss": 0.132,
+      "step": 1907
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.0049447844227266e-05,
+      "loss": 0.1467,
+      "step": 1908
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.002502627924326e-05,
+      "loss": 0.1353,
+      "step": 1909
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.0000609656433838e-05,
+      "loss": 0.1212,
+      "step": 1910
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9976198000054645e-05,
+      "loss": 0.1323,
+      "step": 1911
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9951791334356396e-05,
+      "loss": 0.1307,
+      "step": 1912
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9927389683584867e-05,
+      "loss": 0.1365,
+      "step": 1913
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.99029930719808e-05,
+      "loss": 0.1381,
+      "step": 1914
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9878601523779973e-05,
+      "loss": 0.1343,
+      "step": 1915
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9854215063213142e-05,
+      "loss": 0.1262,
+      "step": 1916
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9829833714505974e-05,
+      "loss": 0.1346,
+      "step": 1917
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9805457501879087e-05,
+      "loss": 0.1366,
+      "step": 1918
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9781086449547976e-05,
+      "loss": 0.1378,
+      "step": 1919
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.975672058172303e-05,
+      "loss": 0.1315,
+      "step": 1920
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.973235992260947e-05,
+      "loss": 0.1285,
+      "step": 1921
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.970800449640735e-05,
+      "loss": 0.1207,
+      "step": 1922
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.968365432731153e-05,
+      "loss": 0.1292,
+      "step": 1923
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9659309439511628e-05,
+      "loss": 0.1422,
+      "step": 1924
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.963496985719205e-05,
+      "loss": 0.141,
+      "step": 1925
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9610635604531898e-05,
+      "loss": 0.1274,
+      "step": 1926
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9586306705705008e-05,
+      "loss": 0.1296,
+      "step": 1927
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9561983184879862e-05,
+      "loss": 0.1222,
+      "step": 1928
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9537665066219635e-05,
+      "loss": 0.1245,
+      "step": 1929
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9513352373882112e-05,
+      "loss": 0.1253,
+      "step": 1930
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9489045132019696e-05,
+      "loss": 0.1328,
+      "step": 1931
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9464743364779385e-05,
+      "loss": 0.1149,
+      "step": 1932
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9440447096302726e-05,
+      "loss": 0.1316,
+      "step": 1933
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9416156350725798e-05,
+      "loss": 0.1348,
+      "step": 1934
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9391871152179204e-05,
+      "loss": 0.1298,
+      "step": 1935
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9367591524788044e-05,
+      "loss": 0.1193,
+      "step": 1936
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9343317492671867e-05,
+      "loss": 0.1256,
+      "step": 1937
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9319049079944674e-05,
+      "loss": 0.1292,
+      "step": 1938
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9294786310714875e-05,
+      "loss": 0.1372,
+      "step": 1939
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.927052920908528e-05,
+      "loss": 0.1261,
+      "step": 1940
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.924627779915308e-05,
+      "loss": 0.1312,
+      "step": 1941
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9222032105009775e-05,
+      "loss": 0.1212,
+      "step": 1942
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9197792150741235e-05,
+      "loss": 0.1297,
+      "step": 1943
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.917355796042759e-05,
+      "loss": 0.139,
+      "step": 1944
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.914932955814327e-05,
+      "loss": 0.1271,
+      "step": 1945
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9125106967956933e-05,
+      "loss": 0.1522,
+      "step": 1946
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.910089021393147e-05,
+      "loss": 0.1429,
+      "step": 1947
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9076679320123996e-05,
+      "loss": 0.141,
+      "step": 1948
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.905247431058578e-05,
+      "loss": 0.1328,
+      "step": 1949
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9028275209362242e-05,
+      "loss": 0.136,
+      "step": 1950
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9004082040492946e-05,
+      "loss": 0.122,
+      "step": 1951
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.897989482801157e-05,
+      "loss": 0.1228,
+      "step": 1952
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.895571359594586e-05,
+      "loss": 0.1333,
+      "step": 1953
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.893153836831762e-05,
+      "loss": 0.1222,
+      "step": 1954
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.89073691691427e-05,
+      "loss": 0.1167,
+      "step": 1955
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.8883206022430956e-05,
+      "loss": 0.1294,
+      "step": 1956
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.8859048952186224e-05,
+      "loss": 0.138,
+      "step": 1957
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.8834897982406312e-05,
+      "loss": 0.1329,
+      "step": 1958
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.8810753137082973e-05,
+      "loss": 0.1313,
+      "step": 1959
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.8786614440201866e-05,
+      "loss": 0.141,
+      "step": 1960
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.876248191574255e-05,
+      "loss": 0.1413,
+      "step": 1961
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.8738355587678442e-05,
+      "loss": 0.1217,
+      "step": 1962
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.8714235479976807e-05,
+      "loss": 0.1648,
+      "step": 1963
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.8690121616598745e-05,
+      "loss": 0.1228,
+      "step": 1964
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.866601402149913e-05,
+      "loss": 0.1318,
+      "step": 1965
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.8641912718626626e-05,
+      "loss": 0.1187,
+      "step": 1966
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.8617817731923633e-05,
+      "loss": 0.1362,
+      "step": 1967
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.8593729085326295e-05,
+      "loss": 0.1344,
+      "step": 1968
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.8569646802764446e-05,
+      "loss": 0.1254,
+      "step": 1969
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.854557090816158e-05,
+      "loss": 0.1415,
+      "step": 1970
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.8521501425434876e-05,
+      "loss": 0.1337,
+      "step": 1971
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.849743837849513e-05,
+      "loss": 0.1371,
+      "step": 1972
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.847338179124675e-05,
+      "loss": 0.1226,
+      "step": 1973
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.844933168758771e-05,
+      "loss": 0.1325,
+      "step": 1974
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.842528809140957e-05,
+      "loss": 0.1282,
+      "step": 1975
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.8401251026597406e-05,
+      "loss": 0.1246,
+      "step": 1976
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.8377220517029804e-05,
+      "loss": 0.1162,
+      "step": 1977
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.8353196586578852e-05,
+      "loss": 0.1291,
+      "step": 1978
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.832917925911009e-05,
+      "loss": 0.1294,
+      "step": 1979
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.8305168558482505e-05,
+      "loss": 0.1307,
+      "step": 1980
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.8281164508548504e-05,
+      "loss": 0.159,
+      "step": 1981
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.825716713315388e-05,
+      "loss": 0.1295,
+      "step": 1982
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.8233176456137797e-05,
+      "loss": 0.1316,
+      "step": 1983
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.8209192501332764e-05,
+      "loss": 0.1256,
+      "step": 1984
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8185215292564612e-05,
+      "loss": 0.1272,
+      "step": 1985
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.816124485365247e-05,
+      "loss": 0.1263,
+      "step": 1986
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8137281208408758e-05,
+      "loss": 0.1398,
+      "step": 1987
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8113324380639117e-05,
+      "loss": 0.1275,
+      "step": 1988
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8089374394142444e-05,
+      "loss": 0.1328,
+      "step": 1989
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8065431272710815e-05,
+      "loss": 0.1196,
+      "step": 1990
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8041495040129507e-05,
+      "loss": 0.1256,
+      "step": 1991
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.801756572017695e-05,
+      "loss": 0.1265,
+      "step": 1992
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7993643336624694e-05,
+      "loss": 0.1354,
+      "step": 1993
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7969727913237404e-05,
+      "loss": 0.1301,
+      "step": 1994
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7945819473772835e-05,
+      "loss": 0.1473,
+      "step": 1995
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.792191804198181e-05,
+      "loss": 0.1322,
+      "step": 1996
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7898023641608186e-05,
+      "loss": 0.1297,
+      "step": 1997
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7874136296388813e-05,
+      "loss": 0.1314,
+      "step": 1998
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7850256030053563e-05,
+      "loss": 0.1241,
+      "step": 1999
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.782638286632526e-05,
+      "loss": 0.1228,
+      "step": 2000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.780251682891968e-05,
+      "loss": 0.1442,
+      "step": 2001
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7778657941545512e-05,
+      "loss": 0.1449,
+      "step": 2002
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7754806227904346e-05,
+      "loss": 0.1403,
+      "step": 2003
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7730961711690655e-05,
+      "loss": 0.1259,
+      "step": 2004
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7707124416591743e-05,
+      "loss": 0.1362,
+      "step": 2005
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7683294366287744e-05,
+      "loss": 0.1315,
+      "step": 2006
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7659471584451615e-05,
+      "loss": 0.1211,
+      "step": 2007
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.763565609474907e-05,
+      "loss": 0.1259,
+      "step": 2008
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.761184792083859e-05,
+      "loss": 0.1323,
+      "step": 2009
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7588047086371384e-05,
+      "loss": 0.121,
+      "step": 2010
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7564253614991384e-05,
+      "loss": 0.134,
+      "step": 2011
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.754046753033517e-05,
+      "loss": 0.1297,
+      "step": 2012
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7516688856032027e-05,
+      "loss": 0.1304,
+      "step": 2013
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7492917615703855e-05,
+      "loss": 0.1275,
+      "step": 2014
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.746915383296518e-05,
+      "loss": 0.1207,
+      "step": 2015
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7445397531423107e-05,
+      "loss": 0.1281,
+      "step": 2016
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7421648734677324e-05,
+      "loss": 0.1353,
+      "step": 2017
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7397907466320044e-05,
+      "loss": 0.1184,
+      "step": 2018
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.737417374993603e-05,
+      "loss": 0.1292,
+      "step": 2019
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.735044760910251e-05,
+      "loss": 0.1399,
+      "step": 2020
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.732672906738922e-05,
+      "loss": 0.1284,
+      "step": 2021
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7303018148358308e-05,
+      "loss": 0.1304,
+      "step": 2022
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.727931487556439e-05,
+      "loss": 0.1209,
+      "step": 2023
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7255619272554457e-05,
+      "loss": 0.1265,
+      "step": 2024
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.72319313628679e-05,
+      "loss": 0.1268,
+      "step": 2025
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.720825117003645e-05,
+      "loss": 0.1278,
+      "step": 2026
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7184578717584176e-05,
+      "loss": 0.1295,
+      "step": 2027
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7160914029027475e-05,
+      "loss": 0.1314,
+      "step": 2028
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7137257127875004e-05,
+      "loss": 0.1141,
+      "step": 2029
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.711360803762771e-05,
+      "loss": 0.1265,
+      "step": 2030
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.708996678177877e-05,
+      "loss": 0.1479,
+      "step": 2031
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.706633338381357e-05,
+      "loss": 0.127,
+      "step": 2032
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7042707867209697e-05,
+      "loss": 0.1116,
+      "step": 2033
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.701909025543691e-05,
+      "loss": 0.1256,
+      "step": 2034
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6995480571957116e-05,
+      "loss": 0.1361,
+      "step": 2035
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6971878840224346e-05,
+      "loss": 0.1322,
+      "step": 2036
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6948285083684725e-05,
+      "loss": 0.1155,
+      "step": 2037
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.692469932577646e-05,
+      "loss": 0.1306,
+      "step": 2038
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6901121589929825e-05,
+      "loss": 0.1442,
+      "step": 2039
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.687755189956709e-05,
+      "loss": 0.1318,
+      "step": 2040
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6853990278102564e-05,
+      "loss": 0.1348,
+      "step": 2041
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6830436748942536e-05,
+      "loss": 0.1303,
+      "step": 2042
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.6806891335485246e-05,
+      "loss": 0.1326,
+      "step": 2043
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.6783354061120878e-05,
+      "loss": 0.1402,
+      "step": 2044
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.675982494923152e-05,
+      "loss": 0.1287,
+      "step": 2045
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.6736304023191178e-05,
+      "loss": 0.1379,
+      "step": 2046
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.67127913063657e-05,
+      "loss": 0.1313,
+      "step": 2047
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.6689286822112785e-05,
+      "loss": 0.1161,
+      "step": 2048
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.666579059378196e-05,
+      "loss": 0.1365,
+      "step": 2049
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.6642302644714537e-05,
+      "loss": 0.1309,
+      "step": 2050
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6618822998243634e-05,
+      "loss": 0.1393,
+      "step": 2051
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6595351677694083e-05,
+      "loss": 0.1396,
+      "step": 2052
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6571888706382473e-05,
+      "loss": 0.123,
+      "step": 2053
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6548434107617073e-05,
+      "loss": 0.1135,
+      "step": 2054
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6524987904697862e-05,
+      "loss": 0.1295,
+      "step": 2055
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6501550120916465e-05,
+      "loss": 0.1253,
+      "step": 2056
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6478120779556133e-05,
+      "loss": 0.1313,
+      "step": 2057
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6454699903891756e-05,
+      "loss": 0.1115,
+      "step": 2058
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.6431287517189797e-05,
+      "loss": 0.1211,
+      "step": 2059
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.6407883642708288e-05,
+      "loss": 0.1326,
+      "step": 2060
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.6384488303696798e-05,
+      "loss": 0.1337,
+      "step": 2061
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.6361101523396438e-05,
+      "loss": 0.1315,
+      "step": 2062
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.6337723325039797e-05,
+      "loss": 0.1281,
+      "step": 2063
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.6314353731850946e-05,
+      "loss": 0.1291,
+      "step": 2064
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.62909927670454e-05,
+      "loss": 0.1287,
+      "step": 2065
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.6267640453830112e-05,
+      "loss": 0.113,
+      "step": 2066
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.6244296815403455e-05,
+      "loss": 0.1349,
+      "step": 2067
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.6220961874955134e-05,
+      "loss": 0.1162,
+      "step": 2068
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.619763565566627e-05,
+      "loss": 0.1359,
+      "step": 2069
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.617431818070929e-05,
+      "loss": 0.1255,
+      "step": 2070
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.6151009473247943e-05,
+      "loss": 0.1323,
+      "step": 2071
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.6127709556437264e-05,
+      "loss": 0.1334,
+      "step": 2072
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.6104418453423554e-05,
+      "loss": 0.1207,
+      "step": 2073
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.6081136187344368e-05,
+      "loss": 0.1464,
+      "step": 2074
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.6057862781328476e-05,
+      "loss": 0.1461,
+      "step": 2075
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.6034598258495843e-05,
+      "loss": 0.1357,
+      "step": 2076
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.6011342641957606e-05,
+      "loss": 0.1249,
+      "step": 2077
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.598809595481608e-05,
+      "loss": 0.1338,
+      "step": 2078
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.596485822016468e-05,
+      "loss": 0.1244,
+      "step": 2079
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5941629461087934e-05,
+      "loss": 0.1165,
+      "step": 2080
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5918409700661467e-05,
+      "loss": 0.124,
+      "step": 2081
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.589519896195195e-05,
+      "loss": 0.1276,
+      "step": 2082
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.58719972680171e-05,
+      "loss": 0.1289,
+      "step": 2083
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.5848804641905634e-05,
+      "loss": 0.1282,
+      "step": 2084
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.5825621106657295e-05,
+      "loss": 0.1383,
+      "step": 2085
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.580244668530276e-05,
+      "loss": 0.1201,
+      "step": 2086
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.5779281400863666e-05,
+      "loss": 0.1212,
+      "step": 2087
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.5756125276352575e-05,
+      "loss": 0.135,
+      "step": 2088
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.5732978334772937e-05,
+      "loss": 0.1407,
+      "step": 2089
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.570984059911911e-05,
+      "loss": 0.1347,
+      "step": 2090
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.568671209237627e-05,
+      "loss": 0.1234,
+      "step": 2091
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.566359283752045e-05,
+      "loss": 0.1514,
+      "step": 2092
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.5640482857518478e-05,
+      "loss": 0.1219,
+      "step": 2093
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.561738217532798e-05,
+      "loss": 0.1236,
+      "step": 2094
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.5594290813897347e-05,
+      "loss": 0.1235,
+      "step": 2095
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.5571208796165677e-05,
+      "loss": 0.1197,
+      "step": 2096
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.554813614506283e-05,
+      "loss": 0.1213,
+      "step": 2097
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.5525072883509345e-05,
+      "loss": 0.1315,
+      "step": 2098
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.5502019034416423e-05,
+      "loss": 0.1302,
+      "step": 2099
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.547897462068592e-05,
+      "loss": 0.1109,
+      "step": 2100
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.545593966521034e-05,
+      "loss": 0.1148,
+      "step": 2101
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.5432914190872757e-05,
+      "loss": 0.126,
+      "step": 2102
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.540989822054685e-05,
+      "loss": 0.134,
+      "step": 2103
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.5386891777096847e-05,
+      "loss": 0.126,
+      "step": 2104
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.5363894883377505e-05,
+      "loss": 0.1188,
+      "step": 2105
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.5340907562234117e-05,
+      "loss": 0.1357,
+      "step": 2106
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.5317929836502448e-05,
+      "loss": 0.1134,
+      "step": 2107
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.5294961729008733e-05,
+      "loss": 0.1183,
+      "step": 2108
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.5272003262569657e-05,
+      "loss": 0.1225,
+      "step": 2109
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.5249054459992335e-05,
+      "loss": 0.1163,
+      "step": 2110
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.5226115344074249e-05,
+      "loss": 0.1235,
+      "step": 2111
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.5203185937603293e-05,
+      "loss": 0.1138,
+      "step": 2112
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.5180266263357712e-05,
+      "loss": 0.1224,
+      "step": 2113
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.5157356344106065e-05,
+      "loss": 0.1249,
+      "step": 2114
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.5134456202607234e-05,
+      "loss": 0.1349,
+      "step": 2115
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.5111565861610378e-05,
+      "loss": 0.1247,
+      "step": 2116
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.5088685343854933e-05,
+      "loss": 0.1252,
+      "step": 2117
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.506581467207057e-05,
+      "loss": 0.1245,
+      "step": 2118
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.5042953868977172e-05,
+      "loss": 0.1256,
+      "step": 2119
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.5020102957284832e-05,
+      "loss": 0.1298,
+      "step": 2120
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.49972619596938e-05,
+      "loss": 0.1364,
+      "step": 2121
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4974430898894498e-05,
+      "loss": 0.1325,
+      "step": 2122
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4951609797567462e-05,
+      "loss": 0.1297,
+      "step": 2123
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4928798678383337e-05,
+      "loss": 0.1373,
+      "step": 2124
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4905997564002849e-05,
+      "loss": 0.1266,
+      "step": 2125
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4883206477076794e-05,
+      "loss": 0.1276,
+      "step": 2126
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4860425440245995e-05,
+      "loss": 0.1264,
+      "step": 2127
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4837654476141297e-05,
+      "loss": 0.1327,
+      "step": 2128
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4814893607383549e-05,
+      "loss": 0.1204,
+      "step": 2129
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4792142856583552e-05,
+      "loss": 0.1335,
+      "step": 2130
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.476940224634207e-05,
+      "loss": 0.1327,
+      "step": 2131
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4746671799249784e-05,
+      "loss": 0.1346,
+      "step": 2132
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4723951537887288e-05,
+      "loss": 0.1254,
+      "step": 2133
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4701241484825052e-05,
+      "loss": 0.1303,
+      "step": 2134
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4678541662623403e-05,
+      "loss": 0.1246,
+      "step": 2135
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.465585209383251e-05,
+      "loss": 0.1221,
+      "step": 2136
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4633172800992345e-05,
+      "loss": 0.1202,
+      "step": 2137
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4610503806632703e-05,
+      "loss": 0.1275,
+      "step": 2138
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4587845133273098e-05,
+      "loss": 0.1378,
+      "step": 2139
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4565196803422826e-05,
+      "loss": 0.1358,
+      "step": 2140
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4542558839580916e-05,
+      "loss": 0.1619,
+      "step": 2141
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4519931264236064e-05,
+      "loss": 0.1297,
+      "step": 2142
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4497314099866671e-05,
+      "loss": 0.1269,
+      "step": 2143
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4474707368940802e-05,
+      "loss": 0.1198,
+      "step": 2144
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4452111093916127e-05,
+      "loss": 0.1201,
+      "step": 2145
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.442952529723996e-05,
+      "loss": 0.1209,
+      "step": 2146
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4406950001349201e-05,
+      "loss": 0.127,
+      "step": 2147
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4384385228670288e-05,
+      "loss": 0.1341,
+      "step": 2148
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4361831001619253e-05,
+      "loss": 0.1168,
+      "step": 2149
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4339287342601604e-05,
+      "loss": 0.1285,
+      "step": 2150
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.431675427401239e-05,
+      "loss": 0.1307,
+      "step": 2151
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4294231818236132e-05,
+      "loss": 0.1283,
+      "step": 2152
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4271719997646765e-05,
+      "loss": 0.1302,
+      "step": 2153
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4249218834607714e-05,
+      "loss": 0.1249,
+      "step": 2154
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4226728351471802e-05,
+      "loss": 0.1386,
+      "step": 2155
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4204248570581219e-05,
+      "loss": 0.1123,
+      "step": 2156
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4181779514267563e-05,
+      "loss": 0.1371,
+      "step": 2157
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4159321204851733e-05,
+      "loss": 0.1364,
+      "step": 2158
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4136873664643985e-05,
+      "loss": 0.1274,
+      "step": 2159
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4114436915943873e-05,
+      "loss": 0.1277,
+      "step": 2160
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.409201098104021e-05,
+      "loss": 0.1129,
+      "step": 2161
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4069595882211087e-05,
+      "loss": 0.1227,
+      "step": 2162
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4047191641723833e-05,
+      "loss": 0.1173,
+      "step": 2163
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4024798281834966e-05,
+      "loss": 0.1275,
+      "step": 2164
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4002415824790227e-05,
+      "loss": 0.1396,
+      "step": 2165
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.3980044292824501e-05,
+      "loss": 0.1315,
+      "step": 2166
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.3957683708161818e-05,
+      "loss": 0.145,
+      "step": 2167
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.393533409301536e-05,
+      "loss": 0.1079,
+      "step": 2168
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.3912995469587378e-05,
+      "loss": 0.1294,
+      "step": 2169
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.3890667860069229e-05,
+      "loss": 0.1218,
+      "step": 2170
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.3868351286641335e-05,
+      "loss": 0.1181,
+      "step": 2171
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.3846045771473115e-05,
+      "loss": 0.1241,
+      "step": 2172
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.382375133672304e-05,
+      "loss": 0.1494,
+      "step": 2173
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3801468004538571e-05,
+      "loss": 0.1356,
+      "step": 2174
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3779195797056107e-05,
+      "loss": 0.126,
+      "step": 2175
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3756934736401044e-05,
+      "loss": 0.1256,
+      "step": 2176
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3734684844687656e-05,
+      "loss": 0.1179,
+      "step": 2177
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3712446144019159e-05,
+      "loss": 0.1123,
+      "step": 2178
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3690218656487643e-05,
+      "loss": 0.1395,
+      "step": 2179
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3668002404174047e-05,
+      "loss": 0.1201,
+      "step": 2180
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3645797409148143e-05,
+      "loss": 0.1249,
+      "step": 2181
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.362360369346855e-05,
+      "loss": 0.1294,
+      "step": 2182
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3601421279182653e-05,
+      "loss": 0.1304,
+      "step": 2183
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3579250188326634e-05,
+      "loss": 0.1276,
+      "step": 2184
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3557090442925397e-05,
+      "loss": 0.1319,
+      "step": 2185
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3534942064992595e-05,
+      "loss": 0.1198,
+      "step": 2186
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.3512805076530607e-05,
+      "loss": 0.119,
+      "step": 2187
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.349067949953045e-05,
+      "loss": 0.1391,
+      "step": 2188
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.346856535597184e-05,
+      "loss": 0.1354,
+      "step": 2189
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.3446462667823139e-05,
+      "loss": 0.1344,
+      "step": 2190
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.3424371457041291e-05,
+      "loss": 0.1166,
+      "step": 2191
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.3402291745571887e-05,
+      "loss": 0.1294,
+      "step": 2192
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.3380223555349053e-05,
+      "loss": 0.1308,
+      "step": 2193
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.3358166908295505e-05,
+      "loss": 0.1131,
+      "step": 2194
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.3336121826322461e-05,
+      "loss": 0.1268,
+      "step": 2195
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.331408833132966e-05,
+      "loss": 0.1237,
+      "step": 2196
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.329206644520534e-05,
+      "loss": 0.1209,
+      "step": 2197
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.3270056189826207e-05,
+      "loss": 0.1275,
+      "step": 2198
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3248057587057397e-05,
+      "loss": 0.1239,
+      "step": 2199
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.322607065875249e-05,
+      "loss": 0.127,
+      "step": 2200
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3204095426753443e-05,
+      "loss": 0.1267,
+      "step": 2201
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.318213191289062e-05,
+      "loss": 0.1224,
+      "step": 2202
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3160180138982736e-05,
+      "loss": 0.1198,
+      "step": 2203
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3138240126836831e-05,
+      "loss": 0.117,
+      "step": 2204
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3116311898248273e-05,
+      "loss": 0.1277,
+      "step": 2205
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3094395475000736e-05,
+      "loss": 0.1166,
+      "step": 2206
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3072490878866131e-05,
+      "loss": 0.1264,
+      "step": 2207
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3050598131604657e-05,
+      "loss": 0.1161,
+      "step": 2208
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3028717254964723e-05,
+      "loss": 0.1371,
+      "step": 2209
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3006848270682931e-05,
+      "loss": 0.1266,
+      "step": 2210
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2984991200484115e-05,
+      "loss": 0.1304,
+      "step": 2211
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2963146066081217e-05,
+      "loss": 0.1215,
+      "step": 2212
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2941312889175366e-05,
+      "loss": 0.1239,
+      "step": 2213
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2919491691455806e-05,
+      "loss": 0.1257,
+      "step": 2214
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.289768249459985e-05,
+      "loss": 0.1336,
+      "step": 2215
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2875885320272933e-05,
+      "loss": 0.1239,
+      "step": 2216
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2854100190128498e-05,
+      "loss": 0.1181,
+      "step": 2217
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2832327125808069e-05,
+      "loss": 0.1182,
+      "step": 2218
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2810566148941166e-05,
+      "loss": 0.1205,
+      "step": 2219
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2788817281145282e-05,
+      "loss": 0.1178,
+      "step": 2220
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2767080544025911e-05,
+      "loss": 0.1203,
+      "step": 2221
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.27453559591765e-05,
+      "loss": 0.1353,
+      "step": 2222
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.272364354817837e-05,
+      "loss": 0.1304,
+      "step": 2223
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2701943332600808e-05,
+      "loss": 0.1218,
+      "step": 2224
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2680255334000964e-05,
+      "loss": 0.1332,
+      "step": 2225
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2658579573923843e-05,
+      "loss": 0.1307,
+      "step": 2226
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2636916073902317e-05,
+      "loss": 0.116,
+      "step": 2227
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2615264855457038e-05,
+      "loss": 0.1147,
+      "step": 2228
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2593625940096494e-05,
+      "loss": 0.1224,
+      "step": 2229
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2571999349316948e-05,
+      "loss": 0.1363,
+      "step": 2230
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2550385104602397e-05,
+      "loss": 0.1242,
+      "step": 2231
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2528783227424599e-05,
+      "loss": 0.126,
+      "step": 2232
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2507193739242996e-05,
+      "loss": 0.1135,
+      "step": 2233
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2485616661504751e-05,
+      "loss": 0.116,
+      "step": 2234
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2464052015644697e-05,
+      "loss": 0.1301,
+      "step": 2235
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2442499823085299e-05,
+      "loss": 0.1319,
+      "step": 2236
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2420960105236651e-05,
+      "loss": 0.1293,
+      "step": 2237
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2399432883496478e-05,
+      "loss": 0.1354,
+      "step": 2238
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2377918179250064e-05,
+      "loss": 0.1176,
+      "step": 2239
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.2356416013870273e-05,
+      "loss": 0.139,
+      "step": 2240
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.2334926408717526e-05,
+      "loss": 0.1132,
+      "step": 2241
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.2313449385139727e-05,
+      "loss": 0.1112,
+      "step": 2242
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.2291984964472327e-05,
+      "loss": 0.1312,
+      "step": 2243
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.2270533168038217e-05,
+      "loss": 0.117,
+      "step": 2244
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.2249094017147775e-05,
+      "loss": 0.1302,
+      "step": 2245
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.2227667533098813e-05,
+      "loss": 0.1254,
+      "step": 2246
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.220625373717654e-05,
+      "loss": 0.1302,
+      "step": 2247
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.2184852650653591e-05,
+      "loss": 0.1319,
+      "step": 2248
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.2163464294789939e-05,
+      "loss": 0.1296,
+      "step": 2249
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.214208869083295e-05,
+      "loss": 0.1326,
+      "step": 2250
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.2120725860017292e-05,
+      "loss": 0.1106,
+      "step": 2251
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.2099375823564949e-05,
+      "loss": 0.1289,
+      "step": 2252
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.2078038602685205e-05,
+      "loss": 0.134,
+      "step": 2253
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.2056714218574618e-05,
+      "loss": 0.1705,
+      "step": 2254
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.203540269241697e-05,
+      "loss": 0.117,
+      "step": 2255
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.201410404538329e-05,
+      "loss": 0.1319,
+      "step": 2256
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1992818298631814e-05,
+      "loss": 0.1178,
+      "step": 2257
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.197154547330794e-05,
+      "loss": 0.1178,
+      "step": 2258
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1950285590544263e-05,
+      "loss": 0.1327,
+      "step": 2259
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1929038671460486e-05,
+      "loss": 0.1291,
+      "step": 2260
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1907804737163456e-05,
+      "loss": 0.1308,
+      "step": 2261
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1886583808747128e-05,
+      "loss": 0.1149,
+      "step": 2262
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1865375907292506e-05,
+      "loss": 0.1175,
+      "step": 2263
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.184418105386769e-05,
+      "loss": 0.1206,
+      "step": 2264
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.182299926952779e-05,
+      "loss": 0.1418,
+      "step": 2265
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.1801830575314932e-05,
+      "loss": 0.1262,
+      "step": 2266
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.1780674992258267e-05,
+      "loss": 0.1263,
+      "step": 2267
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.1759532541373889e-05,
+      "loss": 0.1228,
+      "step": 2268
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.1738403243664864e-05,
+      "loss": 0.1193,
+      "step": 2269
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.1717287120121198e-05,
+      "loss": 0.1437,
+      "step": 2270
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.1696184191719783e-05,
+      "loss": 0.1149,
+      "step": 2271
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.167509447942443e-05,
+      "loss": 0.1438,
+      "step": 2272
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.1654018004185818e-05,
+      "loss": 0.1169,
+      "step": 2273
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.163295478694145e-05,
+      "loss": 0.1328,
+      "step": 2274
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.1611904848615693e-05,
+      "loss": 0.1172,
+      "step": 2275
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.1590868210119692e-05,
+      "loss": 0.1232,
+      "step": 2276
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.1569844892351406e-05,
+      "loss": 0.1128,
+      "step": 2277
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.154883491619556e-05,
+      "loss": 0.1107,
+      "step": 2278
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.1527838302523586e-05,
+      "loss": 0.1444,
+      "step": 2279
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.1506855072193687e-05,
+      "loss": 0.1239,
+      "step": 2280
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.1485885246050762e-05,
+      "loss": 0.1332,
+      "step": 2281
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.1464928844926367e-05,
+      "loss": 0.13,
+      "step": 2282
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.1443985889638764e-05,
+      "loss": 0.1173,
+      "step": 2283
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.1423056400992816e-05,
+      "loss": 0.1204,
+      "step": 2284
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.1402140399780032e-05,
+      "loss": 0.1118,
+      "step": 2285
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.138123790677853e-05,
+      "loss": 0.1203,
+      "step": 2286
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.1360348942752976e-05,
+      "loss": 0.1307,
+      "step": 2287
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.133947352845463e-05,
+      "loss": 0.1194,
+      "step": 2288
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.1318611684621285e-05,
+      "loss": 0.1303,
+      "step": 2289
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.129776343197723e-05,
+      "loss": 0.1173,
+      "step": 2290
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.1276928791233282e-05,
+      "loss": 0.1254,
+      "step": 2291
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.125610778308672e-05,
+      "loss": 0.1193,
+      "step": 2292
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.1235300428221265e-05,
+      "loss": 0.1256,
+      "step": 2293
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.121450674730712e-05,
+      "loss": 0.1255,
+      "step": 2294
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.1193726761000856e-05,
+      "loss": 0.1276,
+      "step": 2295
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.1172960489945466e-05,
+      "loss": 0.1268,
+      "step": 2296
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.1152207954770324e-05,
+      "loss": 0.118,
+      "step": 2297
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.1131469176091128e-05,
+      "loss": 0.1487,
+      "step": 2298
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.1110744174509952e-05,
+      "loss": 0.1266,
+      "step": 2299
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.1090032970615139e-05,
+      "loss": 0.1329,
+      "step": 2300
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.1069335584981355e-05,
+      "loss": 0.1299,
+      "step": 2301
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.1048652038169544e-05,
+      "loss": 0.1209,
+      "step": 2302
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.1027982350726873e-05,
+      "loss": 0.1239,
+      "step": 2303
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.100732654318676e-05,
+      "loss": 0.1073,
+      "step": 2304
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.098668463606885e-05,
+      "loss": 0.1164,
+      "step": 2305
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.0966056649878937e-05,
+      "loss": 0.1204,
+      "step": 2306
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.0945442605109033e-05,
+      "loss": 0.1235,
+      "step": 2307
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.0924842522237267e-05,
+      "loss": 0.1345,
+      "step": 2308
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.0904256421727895e-05,
+      "loss": 0.1217,
+      "step": 2309
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.0883684324031326e-05,
+      "loss": 0.1206,
+      "step": 2310
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.0863126249584001e-05,
+      "loss": 0.115,
+      "step": 2311
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.0842582218808476e-05,
+      "loss": 0.1309,
+      "step": 2312
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.082205225211334e-05,
+      "loss": 0.1316,
+      "step": 2313
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0801536369893194e-05,
+      "loss": 0.1232,
+      "step": 2314
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0781034592528675e-05,
+      "loss": 0.1192,
+      "step": 2315
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.07605469403864e-05,
+      "loss": 0.1221,
+      "step": 2316
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0740073433818934e-05,
+      "loss": 0.1289,
+      "step": 2317
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0719614093164823e-05,
+      "loss": 0.1151,
+      "step": 2318
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0699168938748506e-05,
+      "loss": 0.133,
+      "step": 2319
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0678737990880353e-05,
+      "loss": 0.1192,
+      "step": 2320
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0658321269856633e-05,
+      "loss": 0.1042,
+      "step": 2321
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.0637918795959425e-05,
+      "loss": 0.136,
+      "step": 2322
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.0617530589456715e-05,
+      "loss": 0.1178,
+      "step": 2323
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.0597156670602299e-05,
+      "loss": 0.1233,
+      "step": 2324
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.057679705963576e-05,
+      "loss": 0.1099,
+      "step": 2325
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.0556451776782495e-05,
+      "loss": 0.1187,
+      "step": 2326
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.0536120842253639e-05,
+      "loss": 0.1232,
+      "step": 2327
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.0515804276246094e-05,
+      "loss": 0.1289,
+      "step": 2328
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.0495502098942498e-05,
+      "loss": 0.1246,
+      "step": 2329
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0475214330511154e-05,
+      "loss": 0.1242,
+      "step": 2330
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.045494099110609e-05,
+      "loss": 0.1448,
+      "step": 2331
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0434682100866993e-05,
+      "loss": 0.1191,
+      "step": 2332
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0414437679919179e-05,
+      "loss": 0.107,
+      "step": 2333
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0394207748373611e-05,
+      "loss": 0.1141,
+      "step": 2334
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.037399232632684e-05,
+      "loss": 0.1112,
+      "step": 2335
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0353791433861004e-05,
+      "loss": 0.1305,
+      "step": 2336
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0333605091043835e-05,
+      "loss": 0.1291,
+      "step": 2337
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0313433317928569e-05,
+      "loss": 0.1263,
+      "step": 2338
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0293276134554002e-05,
+      "loss": 0.1389,
+      "step": 2339
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.027313356094443e-05,
+      "loss": 0.1208,
+      "step": 2340
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0253005617109615e-05,
+      "loss": 0.1261,
+      "step": 2341
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0232892323044822e-05,
+      "loss": 0.1287,
+      "step": 2342
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0212793698730713e-05,
+      "loss": 0.1229,
+      "step": 2343
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0192709764133424e-05,
+      "loss": 0.1244,
+      "step": 2344
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0172640539204487e-05,
+      "loss": 0.1267,
+      "step": 2345
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0152586043880794e-05,
+      "loss": 0.1184,
+      "step": 2346
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0132546298084634e-05,
+      "loss": 0.1279,
+      "step": 2347
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0112521321723641e-05,
+      "loss": 0.1169,
+      "step": 2348
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0092511134690768e-05,
+      "loss": 0.1243,
+      "step": 2349
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0072515756864266e-05,
+      "loss": 0.1334,
+      "step": 2350
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0052535208107699e-05,
+      "loss": 0.115,
+      "step": 2351
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0032569508269877e-05,
+      "loss": 0.1171,
+      "step": 2352
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0012618677184885e-05,
+      "loss": 0.115,
+      "step": 2353
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.992682734672005e-06,
+      "loss": 0.1234,
+      "step": 2354
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.972761700535748e-06,
+      "loss": 0.1311,
+      "step": 2355
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.95285559456583e-06,
+      "loss": 0.1187,
+      "step": 2356
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.932964436537099e-06,
+      "loss": 0.1106,
+      "step": 2357
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.913088246209586e-06,
+      "loss": 0.1135,
+      "step": 2358
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.893227043328431e-06,
+      "loss": 0.1283,
+      "step": 2359
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.8733808476239e-06,
+      "loss": 0.1198,
+      "step": 2360
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.853549678811356e-06,
+      "loss": 0.1155,
+      "step": 2361
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.833733556591207e-06,
+      "loss": 0.1269,
+      "step": 2362
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.813932500648947e-06,
+      "loss": 0.1251,
+      "step": 2363
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.794146530655082e-06,
+      "loss": 0.1151,
+      "step": 2364
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.774375666265126e-06,
+      "loss": 0.11,
+      "step": 2365
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.754619927119604e-06,
+      "loss": 0.1175,
+      "step": 2366
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.734879332844019e-06,
+      "loss": 0.1074,
+      "step": 2367
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.715153903048801e-06,
+      "loss": 0.1163,
+      "step": 2368
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.695443657329352e-06,
+      "loss": 0.1149,
+      "step": 2369
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.675748615265957e-06,
+      "loss": 0.1245,
+      "step": 2370
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.656068796423814e-06,
+      "loss": 0.1245,
+      "step": 2371
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.636404220353013e-06,
+      "loss": 0.129,
+      "step": 2372
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.616754906588457e-06,
+      "loss": 0.117,
+      "step": 2373
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.597120874649943e-06,
+      "loss": 0.1234,
+      "step": 2374
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.577502144042035e-06,
+      "loss": 0.1166,
+      "step": 2375
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.557898734254134e-06,
+      "loss": 0.1182,
+      "step": 2376
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.538310664760423e-06,
+      "loss": 0.1192,
+      "step": 2377
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.518737955019796e-06,
+      "loss": 0.1232,
+      "step": 2378
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.499180624475939e-06,
+      "loss": 0.1154,
+      "step": 2379
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.479638692557257e-06,
+      "loss": 0.1292,
+      "step": 2380
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.460112178676825e-06,
+      "loss": 0.1029,
+      "step": 2381
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.440601102232433e-06,
+      "loss": 0.12,
+      "step": 2382
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.42110548260653e-06,
+      "loss": 0.1209,
+      "step": 2383
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.40162533916619e-06,
+      "loss": 0.1285,
+      "step": 2384
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.382160691263141e-06,
+      "loss": 0.1414,
+      "step": 2385
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.362711558233692e-06,
+      "loss": 0.1136,
+      "step": 2386
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.343277959398755e-06,
+      "loss": 0.1355,
+      "step": 2387
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.323859914063814e-06,
+      "loss": 0.1213,
+      "step": 2388
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.304457441518884e-06,
+      "loss": 0.1188,
+      "step": 2389
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.285070561038534e-06,
+      "loss": 0.1177,
+      "step": 2390
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.265699291881818e-06,
+      "loss": 0.126,
+      "step": 2391
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.246343653292291e-06,
+      "loss": 0.1051,
+      "step": 2392
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.227003664497996e-06,
+      "loss": 0.1237,
+      "step": 2393
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.207679344711404e-06,
+      "loss": 0.1256,
+      "step": 2394
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.188370713129435e-06,
+      "loss": 0.1414,
+      "step": 2395
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.169077788933437e-06,
+      "loss": 0.1198,
+      "step": 2396
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.149800591289121e-06,
+      "loss": 0.1341,
+      "step": 2397
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.130539139346596e-06,
+      "loss": 0.1581,
+      "step": 2398
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.111293452240343e-06,
+      "loss": 0.1097,
+      "step": 2399
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.092063549089145e-06,
+      "loss": 0.1138,
+      "step": 2400
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.07284944899614e-06,
+      "loss": 0.1167,
+      "step": 2401
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.053651171048744e-06,
+      "loss": 0.1306,
+      "step": 2402
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.034468734318663e-06,
+      "loss": 0.1137,
+      "step": 2403
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 9.015302157861883e-06,
+      "loss": 0.1414,
+      "step": 2404
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.996151460718604e-06,
+      "loss": 0.1175,
+      "step": 2405
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.977016661913262e-06,
+      "loss": 0.1196,
+      "step": 2406
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.957897780454522e-06,
+      "loss": 0.1203,
+      "step": 2407
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.938794835335194e-06,
+      "loss": 0.1097,
+      "step": 2408
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.919707845532304e-06,
+      "loss": 0.1221,
+      "step": 2409
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.900636830006983e-06,
+      "loss": 0.1137,
+      "step": 2410
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.881581807704528e-06,
+      "loss": 0.106,
+      "step": 2411
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.862542797554341e-06,
+      "loss": 0.1174,
+      "step": 2412
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.843519818469891e-06,
+      "loss": 0.1203,
+      "step": 2413
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.824512889348754e-06,
+      "loss": 0.127,
+      "step": 2414
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.80552202907256e-06,
+      "loss": 0.121,
+      "step": 2415
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.786547256506947e-06,
+      "loss": 0.1315,
+      "step": 2416
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.767588590501605e-06,
+      "loss": 0.1443,
+      "step": 2417
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.748646049890192e-06,
+      "loss": 0.1251,
+      "step": 2418
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.729719653490383e-06,
+      "loss": 0.1191,
+      "step": 2419
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.710809420103789e-06,
+      "loss": 0.1344,
+      "step": 2420
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.69191536851596e-06,
+      "loss": 0.1183,
+      "step": 2421
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.673037517496391e-06,
+      "loss": 0.116,
+      "step": 2422
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.654175885798482e-06,
+      "loss": 0.1253,
+      "step": 2423
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.635330492159499e-06,
+      "loss": 0.1237,
+      "step": 2424
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.616501355300608e-06,
+      "loss": 0.1377,
+      "step": 2425
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.59768849392679e-06,
+      "loss": 0.1195,
+      "step": 2426
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.578891926726886e-06,
+      "loss": 0.1387,
+      "step": 2427
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.56011167237355e-06,
+      "loss": 0.1227,
+      "step": 2428
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.541347749523201e-06,
+      "loss": 0.1154,
+      "step": 2429
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.522600176816068e-06,
+      "loss": 0.1095,
+      "step": 2430
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.503868972876128e-06,
+      "loss": 0.1235,
+      "step": 2431
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.48515415631108e-06,
+      "loss": 0.1131,
+      "step": 2432
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.46645574571237e-06,
+      "loss": 0.1186,
+      "step": 2433
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.447773759655127e-06,
+      "loss": 0.1145,
+      "step": 2434
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.429108216698164e-06,
+      "loss": 0.1318,
+      "step": 2435
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.41045913538398e-06,
+      "loss": 0.1245,
+      "step": 2436
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.391826534238689e-06,
+      "loss": 0.1117,
+      "step": 2437
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.373210431772057e-06,
+      "loss": 0.1294,
+      "step": 2438
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.354610846477467e-06,
+      "loss": 0.1264,
+      "step": 2439
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.33602779683186e-06,
+      "loss": 0.1308,
+      "step": 2440
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.317461301295782e-06,
+      "loss": 0.1178,
+      "step": 2441
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.298911378313328e-06,
+      "loss": 0.1069,
+      "step": 2442
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.28037804631211e-06,
+      "loss": 0.1184,
+      "step": 2443
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.261861323703288e-06,
+      "loss": 0.122,
+      "step": 2444
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.24336122888149e-06,
+      "loss": 0.1182,
+      "step": 2445
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.224877780224857e-06,
+      "loss": 0.1238,
+      "step": 2446
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.206410996094988e-06,
+      "loss": 0.1103,
+      "step": 2447
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.18796089483689e-06,
+      "loss": 0.1214,
+      "step": 2448
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.169527494779042e-06,
+      "loss": 0.122,
+      "step": 2449
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.151110814233326e-06,
+      "loss": 0.1303,
+      "step": 2450
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.132710871494984e-06,
+      "loss": 0.1302,
+      "step": 2451
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.11432768484267e-06,
+      "loss": 0.1199,
+      "step": 2452
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.095961272538358e-06,
+      "loss": 0.1222,
+      "step": 2453
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.077611652827377e-06,
+      "loss": 0.1085,
+      "step": 2454
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.059278843938386e-06,
+      "loss": 0.1138,
+      "step": 2455
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.04096286408331e-06,
+      "loss": 0.1154,
+      "step": 2456
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.022663731457378e-06,
+      "loss": 0.1081,
+      "step": 2457
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.004381464239094e-06,
+      "loss": 0.1345,
+      "step": 2458
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.986116080590176e-06,
+      "loss": 0.1074,
+      "step": 2459
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.967867598655601e-06,
+      "loss": 0.1043,
+      "step": 2460
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.949636036563537e-06,
+      "loss": 0.1295,
+      "step": 2461
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.931421412425339e-06,
+      "loss": 0.1315,
+      "step": 2462
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.913223744335558e-06,
+      "loss": 0.1092,
+      "step": 2463
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.895043050371878e-06,
+      "loss": 0.1265,
+      "step": 2464
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.876879348595134e-06,
+      "loss": 0.116,
+      "step": 2465
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 0.8042522668838501,
+      "eval_runtime": 145.4214,
+      "eval_samples_per_second": 5.24,
+      "eval_steps_per_second": 0.33,
+      "step": 2465
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.858732657049287e-06,
+      "loss": 0.0553,
+      "step": 2466
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.840602993761376e-06,
+      "loss": 0.0481,
+      "step": 2467
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.822490376741554e-06,
+      "loss": 0.0493,
+      "step": 2468
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.804394823983009e-06,
+      "loss": 0.0479,
+      "step": 2469
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.786316353461998e-06,
+      "loss": 0.0439,
+      "step": 2470
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.768254983137815e-06,
+      "loss": 0.0428,
+      "step": 2471
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.750210730952736e-06,
+      "loss": 0.0476,
+      "step": 2472
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.732183614832062e-06,
+      "loss": 0.0486,
+      "step": 2473
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.714173652684064e-06,
+      "loss": 0.0344,
+      "step": 2474
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.696180862399963e-06,
+      "loss": 0.046,
+      "step": 2475
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.678205261853912e-06,
+      "loss": 0.0442,
+      "step": 2476
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.660246868903025e-06,
+      "loss": 0.0366,
+      "step": 2477
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.642305701387279e-06,
+      "loss": 0.0369,
+      "step": 2478
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.624381777129572e-06,
+      "loss": 0.039,
+      "step": 2479
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.606475113935646e-06,
+      "loss": 0.0365,
+      "step": 2480
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.588585729594116e-06,
+      "loss": 0.0393,
+      "step": 2481
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.570713641876434e-06,
+      "loss": 0.0387,
+      "step": 2482
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.5528588685368414e-06,
+      "loss": 0.0338,
+      "step": 2483
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.535021427312417e-06,
+      "loss": 0.035,
+      "step": 2484
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.517201335922986e-06,
+      "loss": 0.0378,
+      "step": 2485
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.4993986120711594e-06,
+      "loss": 0.0381,
+      "step": 2486
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.4816132734423035e-06,
+      "loss": 0.0347,
+      "step": 2487
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.463845337704483e-06,
+      "loss": 0.0351,
+      "step": 2488
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.446094822508509e-06,
+      "loss": 0.0429,
+      "step": 2489
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.428361745487866e-06,
+      "loss": 0.0451,
+      "step": 2490
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.410646124258705e-06,
+      "loss": 0.0377,
+      "step": 2491
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.392947976419867e-06,
+      "loss": 0.0359,
+      "step": 2492
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.3752673195528195e-06,
+      "loss": 0.0433,
+      "step": 2493
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.357604171221644e-06,
+      "loss": 0.0442,
+      "step": 2494
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.33995854897305e-06,
+      "loss": 0.0405,
+      "step": 2495
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.3223304703363135e-06,
+      "loss": 0.0404,
+      "step": 2496
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.304719952823299e-06,
+      "loss": 0.0336,
+      "step": 2497
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.2871270139284325e-06,
+      "loss": 0.0336,
+      "step": 2498
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.2695516711286435e-06,
+      "loss": 0.0358,
+      "step": 2499
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.251993941883428e-06,
+      "loss": 0.0378,
+      "step": 2500
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.234453843634736e-06,
+      "loss": 0.0409,
+      "step": 2501
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.216931393807041e-06,
+      "loss": 0.0379,
+      "step": 2502
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.199426609807275e-06,
+      "loss": 0.0404,
+      "step": 2503
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.1819395090248095e-06,
+      "loss": 0.042,
+      "step": 2504
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.164470108831448e-06,
+      "loss": 0.0309,
+      "step": 2505
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.1470184265814314e-06,
+      "loss": 0.0394,
+      "step": 2506
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.12958447961137e-06,
+      "loss": 0.0428,
+      "step": 2507
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.112168285240287e-06,
+      "loss": 0.0321,
+      "step": 2508
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.094769860769554e-06,
+      "loss": 0.0336,
+      "step": 2509
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.0773892234828805e-06,
+      "loss": 0.0378,
+      "step": 2510
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.060026390646326e-06,
+      "loss": 0.0384,
+      "step": 2511
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.042681379508245e-06,
+      "loss": 0.0353,
+      "step": 2512
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.025354207299301e-06,
+      "loss": 0.0378,
+      "step": 2513
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.008044891232435e-06,
+      "loss": 0.036,
+      "step": 2514
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.990753448502835e-06,
+      "loss": 0.0355,
+      "step": 2515
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.9734798962879575e-06,
+      "loss": 0.037,
+      "step": 2516
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.956224251747459e-06,
+      "loss": 0.0376,
+      "step": 2517
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.938986532023234e-06,
+      "loss": 0.0346,
+      "step": 2518
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.9217667542393536e-06,
+      "loss": 0.0309,
+      "step": 2519
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.904564935502053e-06,
+      "loss": 0.0371,
+      "step": 2520
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.887381092899758e-06,
+      "loss": 0.038,
+      "step": 2521
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.870215243503025e-06,
+      "loss": 0.0337,
+      "step": 2522
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8530674043645185e-06,
+      "loss": 0.0315,
+      "step": 2523
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.835937592519031e-06,
+      "loss": 0.0326,
+      "step": 2524
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.818825824983446e-06,
+      "loss": 0.038,
+      "step": 2525
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8017321187567065e-06,
+      "loss": 0.0337,
+      "step": 2526
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.784656490819832e-06,
+      "loss": 0.0373,
+      "step": 2527
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.767598958135865e-06,
+      "loss": 0.037,
+      "step": 2528
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.750559537649881e-06,
+      "loss": 0.0404,
+      "step": 2529
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.7335382462889755e-06,
+      "loss": 0.0371,
+      "step": 2530
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.716535100962207e-06,
+      "loss": 0.0329,
+      "step": 2531
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.699550118560632e-06,
+      "loss": 0.0396,
+      "step": 2532
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.68258331595725e-06,
+      "loss": 0.0373,
+      "step": 2533
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.665634710006996e-06,
+      "loss": 0.0342,
+      "step": 2534
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.648704317546753e-06,
+      "loss": 0.0396,
+      "step": 2535
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.6317921553952775e-06,
+      "loss": 0.0312,
+      "step": 2536
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.614898240353243e-06,
+      "loss": 0.034,
+      "step": 2537
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.59802258920319e-06,
+      "loss": 0.033,
+      "step": 2538
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.5811652187094956e-06,
+      "loss": 0.0341,
+      "step": 2539
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.564326145618405e-06,
+      "loss": 0.0304,
+      "step": 2540
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.547505386657979e-06,
+      "loss": 0.0357,
+      "step": 2541
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.530702958538062e-06,
+      "loss": 0.0296,
+      "step": 2542
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.513918877950329e-06,
+      "loss": 0.0312,
+      "step": 2543
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.497153161568184e-06,
+      "loss": 0.0359,
+      "step": 2544
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.48040582604682e-06,
+      "loss": 0.0364,
+      "step": 2545
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.463676888023176e-06,
+      "loss": 0.0445,
+      "step": 2546
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.44696636411587e-06,
+      "loss": 0.033,
+      "step": 2547
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.430274270925271e-06,
+      "loss": 0.0393,
+      "step": 2548
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.413600625033428e-06,
+      "loss": 0.0351,
+      "step": 2549
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.396945443004049e-06,
+      "loss": 0.0452,
+      "step": 2550
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.3803087413825234e-06,
+      "loss": 0.035,
+      "step": 2551
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.36369053669586e-06,
+      "loss": 0.0378,
+      "step": 2552
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.347090845452705e-06,
+      "loss": 0.0273,
+      "step": 2553
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.330509684143318e-06,
+      "loss": 0.037,
+      "step": 2554
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.313947069239534e-06,
+      "loss": 0.0411,
+      "step": 2555
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.297403017194778e-06,
+      "loss": 0.0335,
+      "step": 2556
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.280877544444036e-06,
+      "loss": 0.0501,
+      "step": 2557
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.2643706674038154e-06,
+      "loss": 0.0329,
+      "step": 2558
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.2478824024721855e-06,
+      "loss": 0.0383,
+      "step": 2559
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.231412766028697e-06,
+      "loss": 0.0342,
+      "step": 2560
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.214961774434397e-06,
+      "loss": 0.0392,
+      "step": 2561
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.198529444031834e-06,
+      "loss": 0.0393,
+      "step": 2562
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.182115791144988e-06,
+      "loss": 0.032,
+      "step": 2563
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.1657208320793054e-06,
+      "loss": 0.0533,
+      "step": 2564
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.149344583121669e-06,
+      "loss": 0.0353,
+      "step": 2565
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.132987060540341e-06,
+      "loss": 0.0322,
+      "step": 2566
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.116648280585011e-06,
+      "loss": 0.0329,
+      "step": 2567
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.100328259486748e-06,
+      "loss": 0.0355,
+      "step": 2568
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.0840270134579655e-06,
+      "loss": 0.0354,
+      "step": 2569
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.067744558692448e-06,
+      "loss": 0.0326,
+      "step": 2570
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.051480911365293e-06,
+      "loss": 0.0318,
+      "step": 2571
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.035236087632928e-06,
+      "loss": 0.0363,
+      "step": 2572
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.019010103633088e-06,
+      "loss": 0.0347,
+      "step": 2573
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.002802975484773e-06,
+      "loss": 0.037,
+      "step": 2574
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5.986614719288253e-06,
+      "loss": 0.033,
+      "step": 2575
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5.97044535112507e-06,
+      "loss": 0.0313,
+      "step": 2576
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.954294887057982e-06,
+      "loss": 0.0413,
+      "step": 2577
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.938163343130984e-06,
+      "loss": 0.0337,
+      "step": 2578
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.922050735369258e-06,
+      "loss": 0.042,
+      "step": 2579
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.905957079779187e-06,
+      "loss": 0.0344,
+      "step": 2580
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.889882392348334e-06,
+      "loss": 0.0398,
+      "step": 2581
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.873826689045394e-06,
+      "loss": 0.0335,
+      "step": 2582
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.857789985820225e-06,
+      "loss": 0.0325,
+      "step": 2583
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.841772298603806e-06,
+      "loss": 0.0318,
+      "step": 2584
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.825773643308216e-06,
+      "loss": 0.0349,
+      "step": 2585
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.8097940358266435e-06,
+      "loss": 0.0357,
+      "step": 2586
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.793833492033329e-06,
+      "loss": 0.0323,
+      "step": 2587
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.77789202778361e-06,
+      "loss": 0.0335,
+      "step": 2588
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.761969658913843e-06,
+      "loss": 0.0334,
+      "step": 2589
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.746066401241415e-06,
+      "loss": 0.0361,
+      "step": 2590
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.730182270564743e-06,
+      "loss": 0.0363,
+      "step": 2591
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.714317282663245e-06,
+      "loss": 0.0329,
+      "step": 2592
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.6984714532972995e-06,
+      "loss": 0.0345,
+      "step": 2593
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.682644798208281e-06,
+      "loss": 0.0341,
+      "step": 2594
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.666837333118486e-06,
+      "loss": 0.0377,
+      "step": 2595
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.6510490737311735e-06,
+      "loss": 0.0373,
+      "step": 2596
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.6352800357305225e-06,
+      "loss": 0.0326,
+      "step": 2597
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.619530234781589e-06,
+      "loss": 0.0344,
+      "step": 2598
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.603799686530348e-06,
+      "loss": 0.0346,
+      "step": 2599
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.588088406603645e-06,
+      "loss": 0.0338,
+      "step": 2600
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.572396410609168e-06,
+      "loss": 0.0333,
+      "step": 2601
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.556723714135468e-06,
+      "loss": 0.0301,
+      "step": 2602
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.5410703327519075e-06,
+      "loss": 0.0316,
+      "step": 2603
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.525436282008664e-06,
+      "loss": 0.0387,
+      "step": 2604
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.509821577436728e-06,
+      "loss": 0.0316,
+      "step": 2605
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.494226234547842e-06,
+      "loss": 0.0365,
+      "step": 2606
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.478650268834543e-06,
+      "loss": 0.0366,
+      "step": 2607
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.4630936957701104e-06,
+      "loss": 0.038,
+      "step": 2608
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.447556530808545e-06,
+      "loss": 0.033,
+      "step": 2609
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.432038789384592e-06,
+      "loss": 0.0303,
+      "step": 2610
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.4165404869136684e-06,
+      "loss": 0.0346,
+      "step": 2611
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.4010616387919095e-06,
+      "loss": 0.0295,
+      "step": 2612
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.385602260396122e-06,
+      "loss": 0.0352,
+      "step": 2613
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.370162367083745e-06,
+      "loss": 0.0312,
+      "step": 2614
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.354741974192892e-06,
+      "loss": 0.033,
+      "step": 2615
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.339341097042303e-06,
+      "loss": 0.0351,
+      "step": 2616
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.32395975093129e-06,
+      "loss": 0.0362,
+      "step": 2617
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.30859795113981e-06,
+      "loss": 0.0378,
+      "step": 2618
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.293255712928388e-06,
+      "loss": 0.0313,
+      "step": 2619
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.277933051538106e-06,
+      "loss": 0.0462,
+      "step": 2620
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.262629982190612e-06,
+      "loss": 0.0392,
+      "step": 2621
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.247346520088075e-06,
+      "loss": 0.0291,
+      "step": 2622
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.232082680413206e-06,
+      "loss": 0.0348,
+      "step": 2623
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.216838478329214e-06,
+      "loss": 0.035,
+      "step": 2624
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.201613928979788e-06,
+      "loss": 0.0344,
+      "step": 2625
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.186409047489116e-06,
+      "loss": 0.0328,
+      "step": 2626
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.171223848961826e-06,
+      "loss": 0.0332,
+      "step": 2627
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.156058348483006e-06,
+      "loss": 0.038,
+      "step": 2628
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.140912561118183e-06,
+      "loss": 0.0378,
+      "step": 2629
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.125786501913282e-06,
+      "loss": 0.0348,
+      "step": 2630
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.110680185894634e-06,
+      "loss": 0.0294,
+      "step": 2631
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.095593628068973e-06,
+      "loss": 0.0342,
+      "step": 2632
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.0805268434233835e-06,
+      "loss": 0.0315,
+      "step": 2633
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.065479846925319e-06,
+      "loss": 0.0321,
+      "step": 2634
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.050452653522586e-06,
+      "loss": 0.0322,
+      "step": 2635
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.0354452781432886e-06,
+      "loss": 0.0376,
+      "step": 2636
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.02045773569588e-06,
+      "loss": 0.0276,
+      "step": 2637
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.005490041069072e-06,
+      "loss": 0.0329,
+      "step": 2638
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.990542209131891e-06,
+      "loss": 0.0417,
+      "step": 2639
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.975614254733623e-06,
+      "loss": 0.0323,
+      "step": 2640
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.960706192703798e-06,
+      "loss": 0.0334,
+      "step": 2641
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.945818037852201e-06,
+      "loss": 0.0289,
+      "step": 2642
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.930949804968818e-06,
+      "loss": 0.0418,
+      "step": 2643
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.916101508823873e-06,
+      "loss": 0.0383,
+      "step": 2644
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.901273164167763e-06,
+      "loss": 0.0381,
+      "step": 2645
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.886464785731065e-06,
+      "loss": 0.0255,
+      "step": 2646
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.8716763882245355e-06,
+      "loss": 0.0332,
+      "step": 2647
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.856907986339082e-06,
+      "loss": 0.0399,
+      "step": 2648
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.842159594745727e-06,
+      "loss": 0.0323,
+      "step": 2649
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.827431228095633e-06,
+      "loss": 0.0307,
+      "step": 2650
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.812722901020075e-06,
+      "loss": 0.036,
+      "step": 2651
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.798034628130396e-06,
+      "loss": 0.045,
+      "step": 2652
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.7833664240180435e-06,
+      "loss": 0.0299,
+      "step": 2653
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.768718303254504e-06,
+      "loss": 0.0349,
+      "step": 2654
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.754090280391335e-06,
+      "loss": 0.0315,
+      "step": 2655
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.7394823699601245e-06,
+      "loss": 0.0363,
+      "step": 2656
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.724894586472461e-06,
+      "loss": 0.0377,
+      "step": 2657
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.710326944419968e-06,
+      "loss": 0.0374,
+      "step": 2658
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.695779458274238e-06,
+      "loss": 0.0323,
+      "step": 2659
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.681252142486841e-06,
+      "loss": 0.0368,
+      "step": 2660
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.666745011489332e-06,
+      "loss": 0.0331,
+      "step": 2661
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.652258079693183e-06,
+      "loss": 0.0333,
+      "step": 2662
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.637791361489827e-06,
+      "loss": 0.0363,
+      "step": 2663
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.623344871250607e-06,
+      "loss": 0.0326,
+      "step": 2664
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.60891862332676e-06,
+      "loss": 0.0305,
+      "step": 2665
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.59451263204943e-06,
+      "loss": 0.0345,
+      "step": 2666
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.580126911729638e-06,
+      "loss": 0.0337,
+      "step": 2667
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.5657614766582505e-06,
+      "loss": 0.0349,
+      "step": 2668
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.5514163411060045e-06,
+      "loss": 0.0368,
+      "step": 2669
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.53709151932345e-06,
+      "loss": 0.0375,
+      "step": 2670
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.52278702554097e-06,
+      "loss": 0.0349,
+      "step": 2671
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.50850287396877e-06,
+      "loss": 0.0306,
+      "step": 2672
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.4942390787967994e-06,
+      "loss": 0.0312,
+      "step": 2673
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.479995654194827e-06,
+      "loss": 0.0356,
+      "step": 2674
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.46577261431238e-06,
+      "loss": 0.0451,
+      "step": 2675
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.451569973278719e-06,
+      "loss": 0.0387,
+      "step": 2676
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.437387745202856e-06,
+      "loss": 0.0395,
+      "step": 2677
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.423225944173509e-06,
+      "loss": 0.0332,
+      "step": 2678
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.409084584259118e-06,
+      "loss": 0.0282,
+      "step": 2679
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.394963679507821e-06,
+      "loss": 0.0333,
+      "step": 2680
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.3808632439474085e-06,
+      "loss": 0.0283,
+      "step": 2681
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.366783291585363e-06,
+      "loss": 0.0344,
+      "step": 2682
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.352723836408817e-06,
+      "loss": 0.0306,
+      "step": 2683
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.33868489238452e-06,
+      "loss": 0.0388,
+      "step": 2684
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.324666473458874e-06,
+      "loss": 0.0359,
+      "step": 2685
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.31066859355787e-06,
+      "loss": 0.0303,
+      "step": 2686
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.2966912665870965e-06,
+      "loss": 0.0338,
+      "step": 2687
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.2827345064317404e-06,
+      "loss": 0.033,
+      "step": 2688
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.2687983269565354e-06,
+      "loss": 0.0331,
+      "step": 2689
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.2548827420057945e-06,
+      "loss": 0.0259,
+      "step": 2690
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.240987765403356e-06,
+      "loss": 0.0372,
+      "step": 2691
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.227113410952585e-06,
+      "loss": 0.0343,
+      "step": 2692
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.213259692436367e-06,
+      "loss": 0.0302,
+      "step": 2693
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.199426623617092e-06,
+      "loss": 0.0373,
+      "step": 2694
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.185614218236619e-06,
+      "loss": 0.0348,
+      "step": 2695
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.1718224900163025e-06,
+      "loss": 0.0359,
+      "step": 2696
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.1580514526569305e-06,
+      "loss": 0.036,
+      "step": 2697
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.144301119838756e-06,
+      "loss": 0.0315,
+      "step": 2698
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.130571505221467e-06,
+      "loss": 0.0301,
+      "step": 2699
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.116862622444146e-06,
+      "loss": 0.0297,
+      "step": 2700
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.1031744851253105e-06,
+      "loss": 0.0312,
+      "step": 2701
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.089507106862847e-06,
+      "loss": 0.0352,
+      "step": 2702
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.075860501234019e-06,
+      "loss": 0.0311,
+      "step": 2703
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.062234681795479e-06,
+      "loss": 0.0317,
+      "step": 2704
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.048629662083198e-06,
+      "loss": 0.0366,
+      "step": 2705
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.035045455612507e-06,
+      "loss": 0.0296,
+      "step": 2706
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.0214820758780604e-06,
+      "loss": 0.0309,
+      "step": 2707
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.007939536353805e-06,
+      "loss": 0.0316,
+      "step": 2708
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.994417850493004e-06,
+      "loss": 0.0306,
+      "step": 2709
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.980917031728201e-06,
+      "loss": 0.038,
+      "step": 2710
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.967437093471196e-06,
+      "loss": 0.0343,
+      "step": 2711
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.953978049113066e-06,
+      "loss": 0.0347,
+      "step": 2712
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.940539912024113e-06,
+      "loss": 0.0383,
+      "step": 2713
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.927122695553884e-06,
+      "loss": 0.032,
+      "step": 2714
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.913726413031149e-06,
+      "loss": 0.0308,
+      "step": 2715
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.900351077763848e-06,
+      "loss": 0.04,
+      "step": 2716
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.886996703039145e-06,
+      "loss": 0.0371,
+      "step": 2717
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.873663302123381e-06,
+      "loss": 0.0294,
+      "step": 2718
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.860350888262041e-06,
+      "loss": 0.0265,
+      "step": 2719
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.847059474679776e-06,
+      "loss": 0.0347,
+      "step": 2720
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.833789074580371e-06,
+      "loss": 0.0328,
+      "step": 2721
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.820539701146733e-06,
+      "loss": 0.0372,
+      "step": 2722
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.8073113675408972e-06,
+      "loss": 0.0339,
+      "step": 2723
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.7941040869039714e-06,
+      "loss": 0.0353,
+      "step": 2724
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.7809178723561656e-06,
+      "loss": 0.0386,
+      "step": 2725
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.7677527369967712e-06,
+      "loss": 0.0317,
+      "step": 2726
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.754608693904113e-06,
+      "loss": 0.0316,
+      "step": 2727
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.7414857561355906e-06,
+      "loss": 0.0373,
+      "step": 2728
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.728383936727617e-06,
+      "loss": 0.0375,
+      "step": 2729
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.7153032486956285e-06,
+      "loss": 0.0315,
+      "step": 2730
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.7022437050340862e-06,
+      "loss": 0.0347,
+      "step": 2731
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.6892053187164236e-06,
+      "loss": 0.0295,
+      "step": 2732
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.6761881026950717e-06,
+      "loss": 0.0315,
+      "step": 2733
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.663192069901433e-06,
+      "loss": 0.0382,
+      "step": 2734
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.650217233245851e-06,
+      "loss": 0.0325,
+      "step": 2735
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.6372636056176275e-06,
+      "loss": 0.0278,
+      "step": 2736
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.6243311998849783e-06,
+      "loss": 0.0344,
+      "step": 2737
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.6114200288950573e-06,
+      "loss": 0.0317,
+      "step": 2738
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.5985301054739185e-06,
+      "loss": 0.0344,
+      "step": 2739
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.585661442426494e-06,
+      "loss": 0.0301,
+      "step": 2740
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.5728140525366107e-06,
+      "loss": 0.0374,
+      "step": 2741
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.5599879485669616e-06,
+      "loss": 0.0343,
+      "step": 2742
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.5471831432590863e-06,
+      "loss": 0.0337,
+      "step": 2743
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.534399649333364e-06,
+      "loss": 0.0377,
+      "step": 2744
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.5216374794890234e-06,
+      "loss": 0.0354,
+      "step": 2745
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.5088966464040808e-06,
+      "loss": 0.0364,
+      "step": 2746
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.496177162735384e-06,
+      "loss": 0.0316,
+      "step": 2747
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.483479041118548e-06,
+      "loss": 0.033,
+      "step": 2748
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.470802294167985e-06,
+      "loss": 0.0286,
+      "step": 2749
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.458146934476869e-06,
+      "loss": 0.0336,
+      "step": 2750
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.4455129746171157e-06,
+      "loss": 0.034,
+      "step": 2751
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.4329004271394015e-06,
+      "loss": 0.0422,
+      "step": 2752
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.4203093045731105e-06,
+      "loss": 0.0325,
+      "step": 2753
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.4077396194263615e-06,
+      "loss": 0.0308,
+      "step": 2754
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.395191384185975e-06,
+      "loss": 0.0287,
+      "step": 2755
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.382664611317446e-06,
+      "loss": 0.0286,
+      "step": 2756
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.37015931326497e-06,
+      "loss": 0.0311,
+      "step": 2757
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.357675502451396e-06,
+      "loss": 0.035,
+      "step": 2758
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.345213191278229e-06,
+      "loss": 0.0325,
+      "step": 2759
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.332772392125616e-06,
+      "loss": 0.0312,
+      "step": 2760
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3203531173523484e-06,
+      "loss": 0.036,
+      "step": 2761
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.30795537929581e-06,
+      "loss": 0.0372,
+      "step": 2762
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.2955791902720136e-06,
+      "loss": 0.0309,
+      "step": 2763
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.283224562575543e-06,
+      "loss": 0.0327,
+      "step": 2764
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.2708915084795786e-06,
+      "loss": 0.0358,
+      "step": 2765
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.2585800402358703e-06,
+      "loss": 0.0358,
+      "step": 2766
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.246290170074709e-06,
+      "loss": 0.0271,
+      "step": 2767
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.2340219102049514e-06,
+      "loss": 0.0327,
+      "step": 2768
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.221775272813962e-06,
+      "loss": 0.0316,
+      "step": 2769
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.209550270067646e-06,
+      "loss": 0.0288,
+      "step": 2770
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.197346914110419e-06,
+      "loss": 0.0297,
+      "step": 2771
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.1851652170651584e-06,
+      "loss": 0.0381,
+      "step": 2772
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.1730051910332624e-06,
+      "loss": 0.0327,
+      "step": 2773
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.160866848094596e-06,
+      "loss": 0.0297,
+      "step": 2774
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.14875020030746e-06,
+      "loss": 0.0313,
+      "step": 2775
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.1366552597086265e-06,
+      "loss": 0.0299,
+      "step": 2776
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.1245820383133035e-06,
+      "loss": 0.0321,
+      "step": 2777
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.112530548115103e-06,
+      "loss": 0.0346,
+      "step": 2778
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.100500801086073e-06,
+      "loss": 0.0342,
+      "step": 2779
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.088492809176641e-06,
+      "loss": 0.0333,
+      "step": 2780
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.076506584315636e-06,
+      "loss": 0.0359,
+      "step": 2781
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.0645421384102644e-06,
+      "loss": 0.037,
+      "step": 2782
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.052599483346086e-06,
+      "loss": 0.028,
+      "step": 2783
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.040678630987026e-06,
+      "loss": 0.0315,
+      "step": 2784
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.028779593175343e-06,
+      "loss": 0.0304,
+      "step": 2785
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.016902381731618e-06,
+      "loss": 0.0345,
+      "step": 2786
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.0050470084547727e-06,
+      "loss": 0.0318,
+      "step": 2787
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2.9932134851220038e-06,
+      "loss": 0.0331,
+      "step": 2788
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2.9814018234888302e-06,
+      "loss": 0.0294,
+      "step": 2789
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.9696120352890407e-06,
+      "loss": 0.0338,
+      "step": 2790
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.9578441322346856e-06,
+      "loss": 0.0314,
+      "step": 2791
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.9460981260160947e-06,
+      "loss": 0.0326,
+      "step": 2792
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.934374028301834e-06,
+      "loss": 0.0333,
+      "step": 2793
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.922671850738698e-06,
+      "loss": 0.0385,
+      "step": 2794
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.9109916049517264e-06,
+      "loss": 0.0329,
+      "step": 2795
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.8993333025441465e-06,
+      "loss": 0.0337,
+      "step": 2796
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.887696955097405e-06,
+      "loss": 0.0365,
+      "step": 2797
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.876082574171138e-06,
+      "loss": 0.0332,
+      "step": 2798
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2.8644901713031496e-06,
+      "loss": 0.037,
+      "step": 2799
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2.852919758009409e-06,
+      "loss": 0.0334,
+      "step": 2800
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2.8413713457840634e-06,
+      "loss": 0.0328,
+      "step": 2801
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2.829844946099375e-06,
+      "loss": 0.0418,
+      "step": 2802
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2.8183405704057616e-06,
+      "loss": 0.034,
+      "step": 2803
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2.8068582301317425e-06,
+      "loss": 0.0352,
+      "step": 2804
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2.7953979366839606e-06,
+      "loss": 0.0322,
+      "step": 2805
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2.783959701447161e-06,
+      "loss": 0.0342,
+      "step": 2806
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.7725435357841583e-06,
+      "loss": 0.0344,
+      "step": 2807
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.7611494510358605e-06,
+      "loss": 0.0368,
+      "step": 2808
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.749777458521233e-06,
+      "loss": 0.0337,
+      "step": 2809
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.7384275695372885e-06,
+      "loss": 0.0348,
+      "step": 2810
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.727099795359095e-06,
+      "loss": 0.0354,
+      "step": 2811
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.715794147239739e-06,
+      "loss": 0.0339,
+      "step": 2812
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.7045106364103372e-06,
+      "loss": 0.0305,
+      "step": 2813
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.693249274080012e-06,
+      "loss": 0.0294,
+      "step": 2814
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.6820100714358697e-06,
+      "loss": 0.0308,
+      "step": 2815
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.6707930396430244e-06,
+      "loss": 0.0273,
+      "step": 2816
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.6595981898445588e-06,
+      "loss": 0.0265,
+      "step": 2817
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.6484255331615056e-06,
+      "loss": 0.0305,
+      "step": 2818
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.6372750806928746e-06,
+      "loss": 0.0367,
+      "step": 2819
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.6261468435155978e-06,
+      "loss": 0.0317,
+      "step": 2820
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.6150408326845456e-06,
+      "loss": 0.0305,
+      "step": 2821
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.6039570592325145e-06,
+      "loss": 0.0276,
+      "step": 2822
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.5928955341701976e-06,
+      "loss": 0.0294,
+      "step": 2823
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.5818562684861965e-06,
+      "loss": 0.0303,
+      "step": 2824
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.5708392731470022e-06,
+      "loss": 0.0365,
+      "step": 2825
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.5598445590969662e-06,
+      "loss": 0.0301,
+      "step": 2826
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.5488721372583263e-06,
+      "loss": 0.032,
+      "step": 2827
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.5379220185311607e-06,
+      "loss": 0.0361,
+      "step": 2828
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.526994213793393e-06,
+      "loss": 0.0295,
+      "step": 2829
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.516088733900787e-06,
+      "loss": 0.0328,
+      "step": 2830
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.505205589686921e-06,
+      "loss": 0.0343,
+      "step": 2831
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.4943447919631908e-06,
+      "loss": 0.0336,
+      "step": 2832
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.483506351518794e-06,
+      "loss": 0.0329,
+      "step": 2833
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.4726902791207094e-06,
+      "loss": 0.0272,
+      "step": 2834
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.4618965855137012e-06,
+      "loss": 0.0375,
+      "step": 2835
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.4511252814203107e-06,
+      "loss": 0.0313,
+      "step": 2836
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.440376377540815e-06,
+      "loss": 0.0335,
+      "step": 2837
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.429649884553267e-06,
+      "loss": 0.0336,
+      "step": 2838
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.4189458131134295e-06,
+      "loss": 0.03,
+      "step": 2839
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.4082641738548073e-06,
+      "loss": 0.0366,
+      "step": 2840
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.3976049773886288e-06,
+      "loss": 0.0341,
+      "step": 2841
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.3869682343037964e-06,
+      "loss": 0.0344,
+      "step": 2842
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.3763539551669377e-06,
+      "loss": 0.0292,
+      "step": 2843
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.3657621505223576e-06,
+      "loss": 0.0381,
+      "step": 2844
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.3551928308920197e-06,
+      "loss": 0.0342,
+      "step": 2845
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.344646006775575e-06,
+      "loss": 0.0323,
+      "step": 2846
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.3341216886503008e-06,
+      "loss": 0.0358,
+      "step": 2847
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.3236198869711402e-06,
+      "loss": 0.0336,
+      "step": 2848
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.3131406121706554e-06,
+      "loss": 0.0343,
+      "step": 2849
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.302683874659031e-06,
+      "loss": 0.0364,
+      "step": 2850
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.292249684824066e-06,
+      "loss": 0.0314,
+      "step": 2851
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.2818380530311655e-06,
+      "loss": 0.0362,
+      "step": 2852
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.271448989623312e-06,
+      "loss": 0.0305,
+      "step": 2853
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.2610825049210775e-06,
+      "loss": 0.032,
+      "step": 2854
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.250738609222605e-06,
+      "loss": 0.0322,
+      "step": 2855
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.240417312803589e-06,
+      "loss": 0.0342,
+      "step": 2856
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.230118625917285e-06,
+      "loss": 0.035,
+      "step": 2857
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.219842558794477e-06,
+      "loss": 0.0339,
+      "step": 2858
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.209589121643485e-06,
+      "loss": 0.0291,
+      "step": 2859
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.1993583246501555e-06,
+      "loss": 0.0341,
+      "step": 2860
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.189150177977822e-06,
+      "loss": 0.0378,
+      "step": 2861
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.1789646917673434e-06,
+      "loss": 0.0389,
+      "step": 2862
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.1688018761370443e-06,
+      "loss": 0.0361,
+      "step": 2863
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.1586617411827438e-06,
+      "loss": 0.0373,
+      "step": 2864
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.148544296977728e-06,
+      "loss": 0.0295,
+      "step": 2865
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.13844955357273e-06,
+      "loss": 0.0326,
+      "step": 2866
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.1283775209959443e-06,
+      "loss": 0.031,
+      "step": 2867
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.1183282092530065e-06,
+      "loss": 0.0298,
+      "step": 2868
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.1083016283269697e-06,
+      "loss": 0.0324,
+      "step": 2869
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.098297788178308e-06,
+      "loss": 0.0293,
+      "step": 2870
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.088316698744913e-06,
+      "loss": 0.0348,
+      "step": 2871
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.0783583699420658e-06,
+      "loss": 0.0326,
+      "step": 2872
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.0684228116624486e-06,
+      "loss": 0.0343,
+      "step": 2873
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.0585100337761093e-06,
+      "loss": 0.0345,
+      "step": 2874
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.0486200461304726e-06,
+      "loss": 0.0328,
+      "step": 2875
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.038752858550333e-06,
+      "loss": 0.0311,
+      "step": 2876
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.028908480837813e-06,
+      "loss": 0.0355,
+      "step": 2877
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.019086922772398e-06,
+      "loss": 0.0322,
+      "step": 2878
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.009288194110884e-06,
+      "loss": 0.0283,
+      "step": 2879
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.9995123045874057e-06,
+      "loss": 0.0303,
+      "step": 2880
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.9897592639134045e-06,
+      "loss": 0.03,
+      "step": 2881
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.9800290817776123e-06,
+      "loss": 0.0384,
+      "step": 2882
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.97032176784607e-06,
+      "loss": 0.0364,
+      "step": 2883
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.960637331762091e-06,
+      "loss": 0.0315,
+      "step": 2884
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.9509757831462555e-06,
+      "loss": 0.0336,
+      "step": 2885
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.9413371315964214e-06,
+      "loss": 0.0288,
+      "step": 2886
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.9317213866876988e-06,
+      "loss": 0.0312,
+      "step": 2887
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.92212855797243e-06,
+      "loss": 0.0387,
+      "step": 2888
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.9125586549802067e-06,
+      "loss": 0.0313,
+      "step": 2889
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.9030116872178316e-06,
+      "loss": 0.0291,
+      "step": 2890
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.8934876641693355e-06,
+      "loss": 0.0324,
+      "step": 2891
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.883986595295953e-06,
+      "loss": 0.0293,
+      "step": 2892
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.8745084900361087e-06,
+      "loss": 0.0346,
+      "step": 2893
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.8650533578054297e-06,
+      "loss": 0.0295,
+      "step": 2894
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.8556212079966995e-06,
+      "loss": 0.0312,
+      "step": 2895
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.8462120499798896e-06,
+      "loss": 0.0288,
+      "step": 2896
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.8368258931021286e-06,
+      "loss": 0.0245,
+      "step": 2897
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.8274627466876915e-06,
+      "loss": 0.0326,
+      "step": 2898
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.8181226200379847e-06,
+      "loss": 0.0326,
+      "step": 2899
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.8088055224315697e-06,
+      "loss": 0.0323,
+      "step": 2900
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.7995114631241094e-06,
+      "loss": 0.0338,
+      "step": 2901
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.7902404513483935e-06,
+      "loss": 0.0325,
+      "step": 2902
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.7809924963143154e-06,
+      "loss": 0.0314,
+      "step": 2903
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.7717676072088491e-06,
+      "loss": 0.0528,
+      "step": 2904
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.7625657931960777e-06,
+      "loss": 0.0304,
+      "step": 2905
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.753387063417139e-06,
+      "loss": 0.0327,
+      "step": 2906
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.7442314269902504e-06,
+      "loss": 0.0425,
+      "step": 2907
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.7350988930106976e-06,
+      "loss": 0.0332,
+      "step": 2908
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.725989470550793e-06,
+      "loss": 0.028,
+      "step": 2909
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.7169031686599084e-06,
+      "loss": 0.0336,
+      "step": 2910
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.707839996364438e-06,
+      "loss": 0.0313,
+      "step": 2911
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.6987999626678076e-06,
+      "loss": 0.0349,
+      "step": 2912
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.6897830765504468e-06,
+      "loss": 0.0315,
+      "step": 2913
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.680789346969791e-06,
+      "loss": 0.0356,
+      "step": 2914
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.6718187828602822e-06,
+      "loss": 0.0357,
+      "step": 2915
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.6628713931333445e-06,
+      "loss": 0.0287,
+      "step": 2916
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.6539471866773738e-06,
+      "loss": 0.0356,
+      "step": 2917
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.6450461723577392e-06,
+      "loss": 0.0297,
+      "step": 2918
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.6361683590167848e-06,
+      "loss": 0.0371,
+      "step": 2919
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.6273137554737778e-06,
+      "loss": 0.0318,
+      "step": 2920
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.618482370524957e-06,
+      "loss": 0.0373,
+      "step": 2921
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.6096742129434766e-06,
+      "loss": 0.0322,
+      "step": 2922
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.6008892914794238e-06,
+      "loss": 0.0273,
+      "step": 2923
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.592127614859812e-06,
+      "loss": 0.0395,
+      "step": 2924
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.5833891917885396e-06,
+      "loss": 0.0305,
+      "step": 2925
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.5746740309464297e-06,
+      "loss": 0.0335,
+      "step": 2926
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.5659821409911818e-06,
+      "loss": 0.0372,
+      "step": 2927
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.5573135305573748e-06,
+      "loss": 0.0296,
+      "step": 2928
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.5486682082564818e-06,
+      "loss": 0.0263,
+      "step": 2929
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.5400461826768158e-06,
+      "loss": 0.0315,
+      "step": 2930
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.5314474623835673e-06,
+      "loss": 0.0316,
+      "step": 2931
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.5228720559187642e-06,
+      "loss": 0.0312,
+      "step": 2932
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.514319971801273e-06,
+      "loss": 0.0282,
+      "step": 2933
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.5057912185268002e-06,
+      "loss": 0.0327,
+      "step": 2934
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4972858045678716e-06,
+      "loss": 0.0334,
+      "step": 2935
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4888037383738173e-06,
+      "loss": 0.0309,
+      "step": 2936
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.480345028370797e-06,
+      "loss": 0.0322,
+      "step": 2937
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4719096829617418e-06,
+      "loss": 0.0315,
+      "step": 2938
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4634977105263876e-06,
+      "loss": 0.0367,
+      "step": 2939
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.455109119421255e-06,
+      "loss": 0.0339,
+      "step": 2940
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4467439179796205e-06,
+      "loss": 0.0303,
+      "step": 2941
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4384021145115362e-06,
+      "loss": 0.0334,
+      "step": 2942
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4300837173038205e-06,
+      "loss": 0.035,
+      "step": 2943
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4217887346200138e-06,
+      "loss": 0.0323,
+      "step": 2944
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.413517174700424e-06,
+      "loss": 0.0342,
+      "step": 2945
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4052690457620698e-06,
+      "loss": 0.0286,
+      "step": 2946
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.3970443559987029e-06,
+      "loss": 0.0314,
+      "step": 2947
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.3888431135807956e-06,
+      "loss": 0.0278,
+      "step": 2948
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.3806653266555114e-06,
+      "loss": 0.0283,
+      "step": 2949
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.3725110033467297e-06,
+      "loss": 0.0409,
+      "step": 2950
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.3643801517550125e-06,
+      "loss": 0.0272,
+      "step": 2951
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.356272779957604e-06,
+      "loss": 0.0336,
+      "step": 2952
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.3481888960084288e-06,
+      "loss": 0.0373,
+      "step": 2953
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.340128507938071e-06,
+      "loss": 0.0312,
+      "step": 2954
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.3320916237537783e-06,
+      "loss": 0.0302,
+      "step": 2955
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.3240782514394533e-06,
+      "loss": 0.0339,
+      "step": 2956
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.3160883989556338e-06,
+      "loss": 0.0258,
+      "step": 2957
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.3081220742394957e-06,
+      "loss": 0.032,
+      "step": 2958
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.3001792852048473e-06,
+      "loss": 0.0298,
+      "step": 2959
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2922600397421075e-06,
+      "loss": 0.0293,
+      "step": 2960
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.284364345718314e-06,
+      "loss": 0.0335,
+      "step": 2961
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.276492210977112e-06,
+      "loss": 0.0353,
+      "step": 2962
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2686436433387239e-06,
+      "loss": 0.0398,
+      "step": 2963
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2608186505999847e-06,
+      "loss": 0.0288,
+      "step": 2964
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2530172405342876e-06,
+      "loss": 0.0334,
+      "step": 2965
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2452394208916168e-06,
+      "loss": 0.0332,
+      "step": 2966
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2374851993985165e-06,
+      "loss": 0.0325,
+      "step": 2967
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2297545837580805e-06,
+      "loss": 0.0333,
+      "step": 2968
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2220475816499572e-06,
+      "loss": 0.0325,
+      "step": 2969
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.2143642007303418e-06,
+      "loss": 0.0318,
+      "step": 2970
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.2067044486319567e-06,
+      "loss": 0.033,
+      "step": 2971
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.1990683329640567e-06,
+      "loss": 0.0309,
+      "step": 2972
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.19145586131241e-06,
+      "loss": 0.0331,
+      "step": 2973
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.1838670412393033e-06,
+      "loss": 0.0352,
+      "step": 2974
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.1763018802835313e-06,
+      "loss": 0.035,
+      "step": 2975
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.1687603859603686e-06,
+      "loss": 0.0319,
+      "step": 2976
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.1612425657615916e-06,
+      "loss": 0.0337,
+      "step": 2977
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.1537484271554656e-06,
+      "loss": 0.031,
+      "step": 2978
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.14627797758671e-06,
+      "loss": 0.0316,
+      "step": 2979
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.138831224476533e-06,
+      "loss": 0.0323,
+      "step": 2980
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.131408175222587e-06,
+      "loss": 0.0286,
+      "step": 2981
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.1240088371989838e-06,
+      "loss": 0.0325,
+      "step": 2982
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.1166332177562799e-06,
+      "loss": 0.0388,
+      "step": 2983
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.1092813242214618e-06,
+      "loss": 0.0329,
+      "step": 2984
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.1019531638979619e-06,
+      "loss": 0.0328,
+      "step": 2985
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.094648744065624e-06,
+      "loss": 0.0336,
+      "step": 2986
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.08736807198071e-06,
+      "loss": 0.0313,
+      "step": 2987
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.0801111548758957e-06,
+      "loss": 0.0324,
+      "step": 2988
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.0728779999602479e-06,
+      "loss": 0.0346,
+      "step": 2989
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.065668614419238e-06,
+      "loss": 0.0318,
+      "step": 2990
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.058483005414726e-06,
+      "loss": 0.0294,
+      "step": 2991
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.051321180084941e-06,
+      "loss": 0.0333,
+      "step": 2992
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.0441831455444917e-06,
+      "loss": 0.0306,
+      "step": 2993
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.0370689088843615e-06,
+      "loss": 0.0331,
+      "step": 2994
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.0299784771718724e-06,
+      "loss": 0.0341,
+      "step": 2995
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.0229118574507174e-06,
+      "loss": 0.0286,
+      "step": 2996
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.0158690567409256e-06,
+      "loss": 0.0307,
+      "step": 2997
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.0088500820388646e-06,
+      "loss": 0.03,
+      "step": 2998
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.0018549403172345e-06,
+      "loss": 0.0387,
+      "step": 2999
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.948836385250598e-07,
+      "loss": 0.0274,
+      "step": 3000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.879361835876788e-07,
+      "loss": 0.032,
+      "step": 3001
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.810125824067513e-07,
+      "loss": 0.027,
+      "step": 3002
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.741128418602253e-07,
+      "loss": 0.0329,
+      "step": 3003
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.672369688023542e-07,
+      "loss": 0.0334,
+      "step": 3004
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.603849700636768e-07,
+      "loss": 0.0299,
+      "step": 3005
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.535568524510236e-07,
+      "loss": 0.0324,
+      "step": 3006
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.467526227474932e-07,
+      "loss": 0.0333,
+      "step": 3007
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.399722877124511e-07,
+      "loss": 0.0327,
+      "step": 3008
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.3321585408154e-07,
+      "loss": 0.0328,
+      "step": 3009
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.264833285666463e-07,
+      "loss": 0.0506,
+      "step": 3010
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.197747178559035e-07,
+      "loss": 0.0272,
+      "step": 3011
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 9.130900286137001e-07,
+      "loss": 0.0305,
+      "step": 3012
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 9.064292674806518e-07,
+      "loss": 0.0341,
+      "step": 3013
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.997924410736047e-07,
+      "loss": 0.0374,
+      "step": 3014
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.931795559856349e-07,
+      "loss": 0.0328,
+      "step": 3015
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.865906187860262e-07,
+      "loss": 0.0291,
+      "step": 3016
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.800256360202736e-07,
+      "loss": 0.0357,
+      "step": 3017
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.73484614210085e-07,
+      "loss": 0.0362,
+      "step": 3018
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.669675598533517e-07,
+      "loss": 0.0305,
+      "step": 3019
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.604744794241698e-07,
+      "loss": 0.0312,
+      "step": 3020
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.54005379372802e-07,
+      "loss": 0.0299,
+      "step": 3021
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.475602661257048e-07,
+      "loss": 0.0341,
+      "step": 3022
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.411391460855012e-07,
+      "loss": 0.0341,
+      "step": 3023
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.347420256309774e-07,
+      "loss": 0.0309,
+      "step": 3024
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.283689111170751e-07,
+      "loss": 0.0345,
+      "step": 3025
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.220198088748938e-07,
+      "loss": 0.0273,
+      "step": 3026
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.15694725211677e-07,
+      "loss": 0.0338,
+      "step": 3027
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.093936664108071e-07,
+      "loss": 0.0346,
+      "step": 3028
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.031166387318073e-07,
+      "loss": 0.033,
+      "step": 3029
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 7.96863648410312e-07,
+      "loss": 0.0394,
+      "step": 3030
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 7.906347016580967e-07,
+      "loss": 0.0268,
+      "step": 3031
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 7.844298046630339e-07,
+      "loss": 0.0374,
+      "step": 3032
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 7.782489635891149e-07,
+      "loss": 0.0328,
+      "step": 3033
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 7.720921845764367e-07,
+      "loss": 0.03,
+      "step": 3034
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 7.659594737411818e-07,
+      "loss": 0.0305,
+      "step": 3035
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 7.598508371756325e-07,
+      "loss": 0.0338,
+      "step": 3036
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 7.537662809481483e-07,
+      "loss": 0.0324,
+      "step": 3037
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 7.477058111031748e-07,
+      "loss": 0.0359,
+      "step": 3038
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 7.41669433661224e-07,
+      "loss": 0.0288,
+      "step": 3039
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 7.356571546188767e-07,
+      "loss": 0.0366,
+      "step": 3040
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 7.296689799487749e-07,
+      "loss": 0.0313,
+      "step": 3041
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 7.237049155996156e-07,
+      "loss": 0.0309,
+      "step": 3042
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 7.177649674961373e-07,
+      "loss": 0.0314,
+      "step": 3043
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 7.118491415391337e-07,
+      "loss": 0.0307,
+      "step": 3044
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 7.05957443605429e-07,
+      "loss": 0.0294,
+      "step": 3045
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 7.000898795478716e-07,
+      "loss": 0.0299,
+      "step": 3046
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.942464551953492e-07,
+      "loss": 0.0308,
+      "step": 3047
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.884271763527544e-07,
+      "loss": 0.0308,
+      "step": 3048
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.826320488010046e-07,
+      "loss": 0.0325,
+      "step": 3049
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.768610782970253e-07,
+      "loss": 0.0286,
+      "step": 3050
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.711142705737306e-07,
+      "loss": 0.0293,
+      "step": 3051
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.653916313400482e-07,
+      "loss": 0.0317,
+      "step": 3052
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.596931662808836e-07,
+      "loss": 0.0319,
+      "step": 3053
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.540188810571363e-07,
+      "loss": 0.0309,
+      "step": 3054
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.483687813056805e-07,
+      "loss": 0.031,
+      "step": 3055
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.427428726393652e-07,
+      "loss": 0.0303,
+      "step": 3056
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.37141160647009e-07,
+      "loss": 0.0318,
+      "step": 3057
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.315636508933964e-07,
+      "loss": 0.029,
+      "step": 3058
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.260103489192565e-07,
+      "loss": 0.0339,
+      "step": 3059
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.204812602412902e-07,
+      "loss": 0.0302,
+      "step": 3060
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.149763903521288e-07,
+      "loss": 0.0344,
+      "step": 3061
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.094957447203454e-07,
+      "loss": 0.0342,
+      "step": 3062
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.040393287904623e-07,
+      "loss": 0.0354,
+      "step": 3063
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.986071479829131e-07,
+      "loss": 0.0301,
+      "step": 3064
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.931992076940701e-07,
+      "loss": 0.0273,
+      "step": 3065
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.878155132962221e-07,
+      "loss": 0.0297,
+      "step": 3066
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.824560701375603e-07,
+      "loss": 0.0305,
+      "step": 3067
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.771208835422037e-07,
+      "loss": 0.0329,
+      "step": 3068
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.7180995881016e-07,
+      "loss": 0.0301,
+      "step": 3069
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.665233012173421e-07,
+      "loss": 0.0317,
+      "step": 3070
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.612609160155519e-07,
+      "loss": 0.0409,
+      "step": 3071
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.560228084324825e-07,
+      "loss": 0.0288,
+      "step": 3072
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.508089836717051e-07,
+      "loss": 0.035,
+      "step": 3073
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.456194469126735e-07,
+      "loss": 0.0285,
+      "step": 3074
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.404542033107057e-07,
+      "loss": 0.0364,
+      "step": 3075
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.353132579969972e-07,
+      "loss": 0.0324,
+      "step": 3076
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.301966160786015e-07,
+      "loss": 0.0329,
+      "step": 3077
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.251042826384195e-07,
+      "loss": 0.027,
+      "step": 3078
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.200362627352184e-07,
+      "loss": 0.0407,
+      "step": 3079
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.149925614036044e-07,
+      "loss": 0.0317,
+      "step": 3080
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.099731836540251e-07,
+      "loss": 0.0287,
+      "step": 3081
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.049781344727667e-07,
+      "loss": 0.031,
+      "step": 3082
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.000074188219434e-07,
+      "loss": 0.0298,
+      "step": 3083
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.950610416395052e-07,
+      "loss": 0.036,
+      "step": 3084
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.901390078392188e-07,
+      "loss": 0.036,
+      "step": 3085
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.852413223106645e-07,
+      "loss": 0.0335,
+      "step": 3086
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.803679899192392e-07,
+      "loss": 0.0357,
+      "step": 3087
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.755190155061484e-07,
+      "loss": 0.0313,
+      "step": 3088
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.7069440388839423e-07,
+      "loss": 0.0325,
+      "step": 3089
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.658941598587846e-07,
+      "loss": 0.0373,
+      "step": 3090
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.611182881859133e-07,
+      "loss": 0.0305,
+      "step": 3091
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.56366793614163e-07,
+      "loss": 0.0334,
+      "step": 3092
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.5163968086371356e-07,
+      "loss": 0.032,
+      "step": 3093
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.469369546305002e-07,
+      "loss": 0.0332,
+      "step": 3094
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.422586195862527e-07,
+      "loss": 0.0311,
+      "step": 3095
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.376046803784589e-07,
+      "loss": 0.029,
+      "step": 3096
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.3297514163037347e-07,
+      "loss": 0.0299,
+      "step": 3097
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.2837000794102024e-07,
+      "loss": 0.0341,
+      "step": 3098
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.237892838851648e-07,
+      "loss": 0.0358,
+      "step": 3099
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.192329740133366e-07,
+      "loss": 0.0376,
+      "step": 3100
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.1470108285180665e-07,
+      "loss": 0.0257,
+      "step": 3101
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.1019361490258203e-07,
+      "loss": 0.0261,
+      "step": 3102
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.05710574643417e-07,
+      "loss": 0.0331,
+      "step": 3103
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.0125196652780196e-07,
+      "loss": 0.0414,
+      "step": 3104
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.968177949849411e-07,
+      "loss": 0.0325,
+      "step": 3105
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.9240806441977753e-07,
+      "loss": 0.0255,
+      "step": 3106
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.880227792129654e-07,
+      "loss": 0.0369,
+      "step": 3107
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.8366194372088384e-07,
+      "loss": 0.0335,
+      "step": 3108
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.793255622756176e-07,
+      "loss": 0.032,
+      "step": 3109
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.7501363918495127e-07,
+      "loss": 0.0285,
+      "step": 3110
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.7072617873238623e-07,
+      "loss": 0.031,
+      "step": 3111
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.664631851771183e-07,
+      "loss": 0.0368,
+      "step": 3112
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.622246627540293e-07,
+      "loss": 0.0327,
+      "step": 3113
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.5801061567370673e-07,
+      "loss": 0.03,
+      "step": 3114
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.5382104812240745e-07,
+      "loss": 0.0443,
+      "step": 3115
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.4965596426208003e-07,
+      "loss": 0.0337,
+      "step": 3116
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.455153682303508e-07,
+      "loss": 0.0308,
+      "step": 3117
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.4139926414051557e-07,
+      "loss": 0.0305,
+      "step": 3118
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.373076560815397e-07,
+      "loss": 0.0306,
+      "step": 3119
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.332405481180634e-07,
+      "loss": 0.0387,
+      "step": 3120
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.291979442903742e-07,
+      "loss": 0.0322,
+      "step": 3121
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.251798486144264e-07,
+      "loss": 0.0339,
+      "step": 3122
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.2118626508182427e-07,
+      "loss": 0.037,
+      "step": 3123
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1721719765981926e-07,
+      "loss": 0.0296,
+      "step": 3124
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.132726502913186e-07,
+      "loss": 0.0323,
+      "step": 3125
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.0935262689485436e-07,
+      "loss": 0.0298,
+      "step": 3126
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.054571313646143e-07,
+      "loss": 0.0381,
+      "step": 3127
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0158616757041127e-07,
+      "loss": 0.0305,
+      "step": 3128
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.977397393576831e-07,
+      "loss": 0.0284,
+      "step": 3129
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.93917850547501e-07,
+      "loss": 0.0303,
+      "step": 3130
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.901205049365585e-07,
+      "loss": 0.0332,
+      "step": 3131
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.863477062971659e-07,
+      "loss": 0.0333,
+      "step": 3132
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8259945837724735e-07,
+      "loss": 0.0338,
+      "step": 3133
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.7887576490034097e-07,
+      "loss": 0.0313,
+      "step": 3134
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.75176629565585e-07,
+      "loss": 0.0341,
+      "step": 3135
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7150205604773715e-07,
+      "loss": 0.0279,
+      "step": 3136
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.6785204799713583e-07,
+      "loss": 0.0266,
+      "step": 3137
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.642266090397277e-07,
+      "loss": 0.0425,
+      "step": 3138
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.606257427770542e-07,
+      "loss": 0.0312,
+      "step": 3139
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.5704945278623436e-07,
+      "loss": 0.037,
+      "step": 3140
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.534977426199847e-07,
+      "loss": 0.0307,
+      "step": 3141
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.4997061580659953e-07,
+      "loss": 0.0318,
+      "step": 3142
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.4646807584995104e-07,
+      "loss": 0.0325,
+      "step": 3143
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.4299012622948924e-07,
+      "loss": 0.0405,
+      "step": 3144
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.3953677040023103e-07,
+      "loss": 0.0274,
+      "step": 3145
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.361080117927683e-07,
+      "loss": 0.0298,
+      "step": 3146
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.3270385381325143e-07,
+      "loss": 0.0332,
+      "step": 3147
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.2932429984339753e-07,
+      "loss": 0.0337,
+      "step": 3148
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.2596935324047942e-07,
+      "loss": 0.0338,
+      "step": 3149
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.2263901733732552e-07,
+      "loss": 0.0338,
+      "step": 3150
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.193332954423144e-07,
+      "loss": 0.0303,
+      "step": 3151
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.16052190839372e-07,
+      "loss": 0.032,
+      "step": 3152
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.1279570678797433e-07,
+      "loss": 0.0332,
+      "step": 3153
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.0956384652313642e-07,
+      "loss": 0.0328,
+      "step": 3154
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.0635661325540955e-07,
+      "loss": 0.0278,
+      "step": 3155
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.0317401017088122e-07,
+      "loss": 0.0342,
+      "step": 3156
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.0001604043117516e-07,
+      "loss": 0.029,
+      "step": 3157
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.9688270717343749e-07,
+      "loss": 0.0365,
+      "step": 3158
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.9377401351034775e-07,
+      "loss": 0.0355,
+      "step": 3159
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.906899625301023e-07,
+      "loss": 0.0243,
+      "step": 3160
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.8763055729641987e-07,
+      "loss": 0.0427,
+      "step": 3161
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.8459580084853877e-07,
+      "loss": 0.0313,
+      "step": 3162
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.8158569620120024e-07,
+      "loss": 0.0311,
+      "step": 3163
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.7860024634467064e-07,
+      "loss": 0.0351,
+      "step": 3164
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.756394542447165e-07,
+      "loss": 0.0333,
+      "step": 3165
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.7270332284260726e-07,
+      "loss": 0.0278,
+      "step": 3166
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.6979185505511807e-07,
+      "loss": 0.032,
+      "step": 3167
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.6690505377452148e-07,
+      "loss": 0.0346,
+      "step": 3168
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.6404292186858738e-07,
+      "loss": 0.0322,
+      "step": 3169
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.6120546218057753e-07,
+      "loss": 0.0267,
+      "step": 3170
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.5839267752924548e-07,
+      "loss": 0.0278,
+      "step": 3171
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.5560457070883105e-07,
+      "loss": 0.0312,
+      "step": 3172
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.5284114448906317e-07,
+      "loss": 0.0374,
+      "step": 3173
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.5010240161514034e-07,
+      "loss": 0.0276,
+      "step": 3174
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.4738834480775844e-07,
+      "loss": 0.0343,
+      "step": 3175
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.4469897676307743e-07,
+      "loss": 0.0327,
+      "step": 3176
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.4203430015273244e-07,
+      "loss": 0.0319,
+      "step": 3177
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.3939431762383659e-07,
+      "loss": 0.0355,
+      "step": 3178
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.36779031798967e-07,
+      "loss": 0.0329,
+      "step": 3179
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.3418844527616213e-07,
+      "loss": 0.0345,
+      "step": 3180
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.3162256062893563e-07,
+      "loss": 0.0272,
+      "step": 3181
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2908138040625407e-07,
+      "loss": 0.0299,
+      "step": 3182
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2656490713253977e-07,
+      "loss": 0.0295,
+      "step": 3183
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2407314330768193e-07,
+      "loss": 0.0267,
+      "step": 3184
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2160609140701164e-07,
+      "loss": 0.0307,
+      "step": 3185
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1916375388131839e-07,
+      "loss": 0.0363,
+      "step": 3186
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1674613315683647e-07,
+      "loss": 0.0305,
+      "step": 3187
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1435323163525025e-07,
+      "loss": 0.0321,
+      "step": 3188
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1198505169368322e-07,
+      "loss": 0.0355,
+      "step": 3189
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.0964159568470079e-07,
+      "loss": 0.0372,
+      "step": 3190
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.0732286593631014e-07,
+      "loss": 0.0295,
+      "step": 3191
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.0502886475195484e-07,
+      "loss": 0.0332,
+      "step": 3192
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.0275959441051475e-07,
+      "loss": 0.0398,
+      "step": 3193
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.0051505716629217e-07,
+      "loss": 0.0221,
+      "step": 3194
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 9.829525524903405e-08,
+      "loss": 0.0287,
+      "step": 3195
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 9.610019086390421e-08,
+      "loss": 0.0323,
+      "step": 3196
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 9.392986619149447e-08,
+      "loss": 0.0263,
+      "step": 3197
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 9.178428338781908e-08,
+      "loss": 0.0323,
+      "step": 3198
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.96634445843203e-08,
+      "loss": 0.048,
+      "step": 3199
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.756735188785448e-08,
+      "loss": 0.0292,
+      "step": 3200
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.549600738068931e-08,
+      "loss": 0.0271,
+      "step": 3201
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.344941312051769e-08,
+      "loss": 0.0295,
+      "step": 3202
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.142757114044386e-08,
+      "loss": 0.027,
+      "step": 3203
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.94304834489723e-08,
+      "loss": 0.0337,
+      "step": 3204
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.74581520300327e-08,
+      "loss": 0.0296,
+      "step": 3205
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.551057884294666e-08,
+      "loss": 0.0313,
+      "step": 3206
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.358776582244986e-08,
+      "loss": 0.0319,
+      "step": 3207
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.168971487868103e-08,
+      "loss": 0.0286,
+      "step": 3208
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.98164278971708e-08,
+      "loss": 0.0353,
+      "step": 3209
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.796790673886388e-08,
+      "loss": 0.0273,
+      "step": 3210
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.61441532400886e-08,
+      "loss": 0.031,
+      "step": 3211
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.434516921257905e-08,
+      "loss": 0.0291,
+      "step": 3212
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.257095644345568e-08,
+      "loss": 0.0369,
+      "step": 3213
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.082151669523361e-08,
+      "loss": 0.03,
+      "step": 3214
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.909685170582269e-08,
+      "loss": 0.031,
+      "step": 3215
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.739696318851629e-08,
+      "loss": 0.0314,
+      "step": 3216
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.572185283199138e-08,
+      "loss": 0.0278,
+      "step": 3217
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.4071522300319643e-08,
+      "loss": 0.0319,
+      "step": 3218
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.244597323295075e-08,
+      "loss": 0.0307,
+      "step": 3219
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.0845207244715196e-08,
+      "loss": 0.0322,
+      "step": 3220
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.9269225925827054e-08,
+      "loss": 0.0272,
+      "step": 3221
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.771803084187565e-08,
+      "loss": 0.0312,
+      "step": 3222
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.619162353382833e-08,
+      "loss": 0.0343,
+      "step": 3223
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.4690005518033264e-08,
+      "loss": 0.0292,
+      "step": 3224
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.3213178286202747e-08,
+      "loss": 0.029,
+      "step": 3225
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.1761143305432684e-08,
+      "loss": 0.0324,
+      "step": 3226
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.0333902018185874e-08,
+      "loss": 0.0446,
+      "step": 3227
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.89314558422893e-08,
+      "loss": 0.0319,
+      "step": 3228
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.75538061709424e-08,
+      "loss": 0.0285,
+      "step": 3229
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.620095437271986e-08,
+      "loss": 0.0315,
+      "step": 3230
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.487290179154667e-08,
+      "loss": 0.0293,
+      "step": 3231
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.356964974672305e-08,
+      "loss": 0.0291,
+      "step": 3232
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.229119953290782e-08,
+      "loss": 0.028,
+      "step": 3233
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.103755242012396e-08,
+      "loss": 0.0291,
+      "step": 3234
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.980870965375582e-08,
+      "loss": 0.0356,
+      "step": 3235
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.8604672454538018e-08,
+      "loss": 0.0269,
+      "step": 3236
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.742544201857211e-08,
+      "loss": 0.0386,
+      "step": 3237
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.6271019517318252e-08,
+      "loss": 0.03,
+      "step": 3238
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.5141406097584085e-08,
+      "loss": 0.0315,
+      "step": 3239
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.403660288153864e-08,
+      "loss": 0.0351,
+      "step": 3240
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2956610966695657e-08,
+      "loss": 0.0267,
+      "step": 3241
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.190143142593304e-08,
+      "loss": 0.0303,
+      "step": 3242
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.0871065307470627e-08,
+      "loss": 0.0291,
+      "step": 3243
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.9865513634884094e-08,
+      "loss": 0.0303,
+      "step": 3244
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.8884777407091047e-08,
+      "loss": 0.0377,
+      "step": 3245
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.7928857598364934e-08,
+      "loss": 0.0237,
+      "step": 3246
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.6997755158323913e-08,
+      "loss": 0.0286,
+      "step": 3247
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.609147101193087e-08,
+      "loss": 0.0345,
+      "step": 3248
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.5210006059496184e-08,
+      "loss": 0.0324,
+      "step": 3249
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.4353361176669411e-08,
+      "loss": 0.0293,
+      "step": 3250
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.352153721445315e-08,
+      "loss": 0.0328,
+      "step": 3251
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2714534999183625e-08,
+      "loss": 0.0399,
+      "step": 3252
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.1932355332544554e-08,
+      "loss": 0.0313,
+      "step": 3253
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.1174998991556051e-08,
+      "loss": 0.0359,
+      "step": 3254
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.044246672858573e-08,
+      "loss": 0.0287,
+      "step": 3255
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 9.734759271334825e-09,
+      "loss": 0.0339,
+      "step": 3256
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 9.051877322840963e-09,
+      "loss": 0.0301,
+      "step": 3257
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.393821561489268e-09,
+      "loss": 0.0339,
+      "step": 3258
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 7.760592640992937e-09,
+      "loss": 0.0299,
+      "step": 3259
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 7.1521911904098846e-09,
+      "loss": 0.0381,
+      "step": 3260
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.568617814128875e-09,
+      "loss": 0.0307,
+      "step": 3261
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.009873091875062e-09,
+      "loss": 0.031,
+      "step": 3262
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.475957578712776e-09,
+      "loss": 0.0291,
+      "step": 3263
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.966871805031637e-09,
+      "loss": 0.0319,
+      "step": 3264
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.482616276565988e-09,
+      "loss": 0.0353,
+      "step": 3265
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.023191474375465e-09,
+      "loss": 0.0331,
+      "step": 3266
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.588597854858877e-09,
+      "loss": 0.0282,
+      "step": 3267
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.1788358497431005e-09,
+      "loss": 0.0352,
+      "step": 3268
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.793905866088631e-09,
+      "loss": 0.025,
+      "step": 3269
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.433808286289585e-09,
+      "loss": 0.0341,
+      "step": 3270
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0985434680653726e-09,
+      "loss": 0.0317,
+      "step": 3271
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.7881117444745742e-09,
+      "loss": 0.0268,
+      "step": 3272
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.5025134239010642e-09,
+      "loss": 0.0251,
+      "step": 3273
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2417487900595603e-09,
+      "loss": 0.0244,
+      "step": 3274
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.005818101995626e-09,
+      "loss": 0.0325,
+      "step": 3275
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.947215940828922e-10,
+      "loss": 0.0321,
+      "step": 3276
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.084594760313866e-10,
+      "loss": 0.0283,
+      "step": 3277
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.470319328708783e-10,
+      "loss": 0.0301,
+      "step": 3278
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.1043912496475647e-10,
+      "loss": 0.0325,
+      "step": 3279
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.986811880044792e-10,
+      "loss": 0.0314,
+      "step": 3280
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.1175823301512456e-10,
+      "loss": 0.0335,
+      "step": 3281
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.967034634428824e-11,
+      "loss": 0.0325,
+      "step": 3282
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2417589670410401e-11,
+      "loss": 0.0365,
+      "step": 3283
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0,
+      "loss": 0.0414,
+      "step": 3284
+    },
+    {
+      "epoch": 4.0,
+      "eval_loss": 1.034638524055481,
+      "eval_runtime": 143.4457,
+      "eval_samples_per_second": 5.312,
+      "eval_steps_per_second": 0.335,
+      "step": 3284
+    },
+    {
+      "epoch": 4.0,
+      "step": 3284,
+      "total_flos": 8.701459909683708e+18,
+      "train_loss": 0.21852872384163063,
+      "train_runtime": 175618.5328,
+      "train_samples_per_second": 1.198,
+      "train_steps_per_second": 0.019
+    }
+  ],
+  "max_steps": 3284,
+  "num_train_epochs": 4,
+  "total_flos": 8.701459909683708e+18,
+  "trial_name": null,
+  "trial_params": null
+}