diff --git "a/checkpoint-10645/trainer_state.json" "b/checkpoint-10645/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-10645/trainer_state.json"
@@ -0,0 +1,63931 @@
+{
+  "best_metric": 0.3247721791267395,
+  "best_model_checkpoint": "./medium-article-titles-engagement/checkpoint-8516",
+  "epoch": 5.0,
+  "global_step": 10645,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.6818,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.6986,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-06,
+      "loss": 0.701,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.7021,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 0.6901,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-06,
+      "loss": 0.6941,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 0.6989,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.7072,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-06,
+      "loss": 0.6891,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 0.6904,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.6864,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2e-05,
+      "loss": 0.6989,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 0.7111,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 0.6903,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5e-05,
+      "loss": 0.6937,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.6822,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 0.6945,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8e-05,
+      "loss": 0.6798,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9e-05,
+      "loss": 0.6981,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.6893,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1e-05,
+      "loss": 0.6848,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 0.6884,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 0.6881,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4e-05,
+      "loss": 0.6917,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-05,
+      "loss": 0.6933,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 0.6853,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 0.6998,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.673,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9e-05,
+      "loss": 0.6826,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-05,
+      "loss": 0.6645,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1e-05,
+      "loss": 0.6649,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.6866,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3e-05,
+      "loss": 0.7105,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 0.6383,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5e-05,
+      "loss": 0.6576,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6e-05,
+      "loss": 0.6313,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7e-05,
+      "loss": 0.6826,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8e-05,
+      "loss": 0.6783,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 0.6609,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4e-05,
+      "loss": 0.6377,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1e-05,
+      "loss": 0.7075,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2e-05,
+      "loss": 0.6811,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3e-05,
+      "loss": 0.6503,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.6835,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5e-05,
+      "loss": 0.6662,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 0.6585,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7e-05,
+      "loss": 0.6833,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8e-05,
+      "loss": 0.652,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9e-05,
+      "loss": 0.6244,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-05,
+      "loss": 0.7156,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999764595103578e-05,
+      "loss": 0.6766,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999529190207157e-05,
+      "loss": 0.6881,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999293785310735e-05,
+      "loss": 0.6303,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9990583804143125e-05,
+      "loss": 0.6781,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998822975517891e-05,
+      "loss": 0.6836,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998587570621469e-05,
+      "loss": 0.6516,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998352165725047e-05,
+      "loss": 0.6276,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9981167608286253e-05,
+      "loss": 0.6594,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997881355932204e-05,
+      "loss": 0.6584,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997645951035782e-05,
+      "loss": 0.6945,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99741054613936e-05,
+      "loss": 0.7133,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997175141242938e-05,
+      "loss": 0.6994,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9969397363465166e-05,
+      "loss": 0.6234,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9967043314500943e-05,
+      "loss": 0.7266,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996468926553672e-05,
+      "loss": 0.678,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996233521657251e-05,
+      "loss": 0.684,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995998116760829e-05,
+      "loss": 0.6161,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9957627118644066e-05,
+      "loss": 0.6386,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9955273069679856e-05,
+      "loss": 0.6564,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9952919020715633e-05,
+      "loss": 0.71,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995056497175141e-05,
+      "loss": 0.6855,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9948210922787195e-05,
+      "loss": 0.6838,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994585687382298e-05,
+      "loss": 0.6483,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994350282485876e-05,
+      "loss": 0.7043,
+      "step": 74
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994114877589454e-05,
+      "loss": 0.6453,
+      "step": 75
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9938794726930323e-05,
+      "loss": 0.6593,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993644067796611e-05,
+      "loss": 0.6753,
+      "step": 77
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9934086629001885e-05,
+      "loss": 0.783,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993173258003767e-05,
+      "loss": 0.661,
+      "step": 79
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992937853107345e-05,
+      "loss": 0.6719,
+      "step": 80
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992702448210923e-05,
+      "loss": 0.7038,
+      "step": 81
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992467043314501e-05,
+      "loss": 0.7013,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99223163841808e-05,
+      "loss": 0.7031,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9919962335216575e-05,
+      "loss": 0.6391,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991760828625235e-05,
+      "loss": 0.7058,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991525423728814e-05,
+      "loss": 0.7178,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991290018832392e-05,
+      "loss": 0.6847,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99105461393597e-05,
+      "loss": 0.6623,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.990819209039548e-05,
+      "loss": 0.6607,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9905838041431265e-05,
+      "loss": 0.6847,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.990348399246705e-05,
+      "loss": 0.6691,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9901129943502826e-05,
+      "loss": 0.649,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.989877589453861e-05,
+      "loss": 0.7197,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9896421845574393e-05,
+      "loss": 0.6891,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.989406779661017e-05,
+      "loss": 0.648,
+      "step": 95
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9891713747645955e-05,
+      "loss": 0.6452,
+      "step": 96
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988935969868174e-05,
+      "loss": 0.6559,
+      "step": 97
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9887005649717516e-05,
+      "loss": 0.6436,
+      "step": 98
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988465160075329e-05,
+      "loss": 0.6379,
+      "step": 99
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9882297551789083e-05,
+      "loss": 0.6937,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987994350282486e-05,
+      "loss": 0.613,
+      "step": 101
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987758945386064e-05,
+      "loss": 0.6465,
+      "step": 102
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987523540489643e-05,
+      "loss": 0.6608,
+      "step": 103
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9872881355932206e-05,
+      "loss": 0.6617,
+      "step": 104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987052730696798e-05,
+      "loss": 0.6937,
+      "step": 105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.986817325800377e-05,
+      "loss": 0.6472,
+      "step": 106
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.986581920903955e-05,
+      "loss": 0.6862,
+      "step": 107
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9863465160075334e-05,
+      "loss": 0.6754,
+      "step": 108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.986111111111111e-05,
+      "loss": 0.6848,
+      "step": 109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9858757062146896e-05,
+      "loss": 0.6845,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.985640301318268e-05,
+      "loss": 0.631,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.985404896421846e-05,
+      "loss": 0.5851,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.985169491525424e-05,
+      "loss": 0.6098,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9849340866290024e-05,
+      "loss": 0.7105,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.98469868173258e-05,
+      "loss": 0.6077,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.984463276836158e-05,
+      "loss": 0.7026,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.984227871939737e-05,
+      "loss": 0.6297,
+      "step": 117
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.983992467043315e-05,
+      "loss": 0.7316,
+      "step": 118
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9837570621468924e-05,
+      "loss": 0.6758,
+      "step": 119
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9835216572504714e-05,
+      "loss": 0.6324,
+      "step": 120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.983286252354049e-05,
+      "loss": 0.719,
+      "step": 121
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9830508474576276e-05,
+      "loss": 0.6764,
+      "step": 122
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.982815442561205e-05,
+      "loss": 0.7366,
+      "step": 123
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.982580037664784e-05,
+      "loss": 0.6851,
+      "step": 124
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.982344632768362e-05,
+      "loss": 0.6576,
+      "step": 125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.98210922787194e-05,
+      "loss": 0.7361,
+      "step": 126
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.981873822975518e-05,
+      "loss": 0.6496,
+      "step": 127
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9816384180790966e-05,
+      "loss": 0.68,
+      "step": 128
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.981403013182674e-05,
+      "loss": 0.6689,
+      "step": 129
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.981167608286253e-05,
+      "loss": 0.6733,
+      "step": 130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980932203389831e-05,
+      "loss": 0.6602,
+      "step": 131
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980696798493409e-05,
+      "loss": 0.6509,
+      "step": 132
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980461393596987e-05,
+      "loss": 0.6294,
+      "step": 133
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9802259887005656e-05,
+      "loss": 0.6351,
+      "step": 134
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.979990583804143e-05,
+      "loss": 0.6601,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.979755178907721e-05,
+      "loss": 0.687,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9795197740113e-05,
+      "loss": 0.6701,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.979284369114878e-05,
+      "loss": 0.6944,
+      "step": 138
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.979048964218456e-05,
+      "loss": 0.6964,
+      "step": 139
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.978813559322034e-05,
+      "loss": 0.625,
+      "step": 140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.978578154425612e-05,
+      "loss": 0.6573,
+      "step": 141
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9783427495291907e-05,
+      "loss": 0.6693,
+      "step": 142
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9781073446327684e-05,
+      "loss": 0.7238,
+      "step": 143
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.977871939736347e-05,
+      "loss": 0.6496,
+      "step": 144
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.977636534839925e-05,
+      "loss": 0.6629,
+      "step": 145
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.977401129943503e-05,
+      "loss": 0.6345,
+      "step": 146
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.977165725047081e-05,
+      "loss": 0.6223,
+      "step": 147
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9769303201506597e-05,
+      "loss": 0.6877,
+      "step": 148
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9766949152542374e-05,
+      "loss": 0.6346,
+      "step": 149
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.976459510357816e-05,
+      "loss": 0.6589,
+      "step": 150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.976224105461394e-05,
+      "loss": 0.6047,
+      "step": 151
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.975988700564972e-05,
+      "loss": 0.7097,
+      "step": 152
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.97575329566855e-05,
+      "loss": 0.6549,
+      "step": 153
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9755178907721287e-05,
+      "loss": 0.6175,
+      "step": 154
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9752824858757064e-05,
+      "loss": 0.6594,
+      "step": 155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.975047080979285e-05,
+      "loss": 0.6309,
+      "step": 156
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9748116760828625e-05,
+      "loss": 0.6301,
+      "step": 157
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.974576271186441e-05,
+      "loss": 0.6198,
+      "step": 158
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.974340866290019e-05,
+      "loss": 0.6832,
+      "step": 159
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.974105461393597e-05,
+      "loss": 0.7371,
+      "step": 160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9738700564971754e-05,
+      "loss": 0.5944,
+      "step": 161
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.973634651600754e-05,
+      "loss": 0.6685,
+      "step": 162
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9733992467043315e-05,
+      "loss": 0.665,
+      "step": 163
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.97316384180791e-05,
+      "loss": 0.5687,
+      "step": 164
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.972928436911488e-05,
+      "loss": 0.6496,
+      "step": 165
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.972693032015066e-05,
+      "loss": 0.6105,
+      "step": 166
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9724576271186444e-05,
+      "loss": 0.6612,
+      "step": 167
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.972222222222223e-05,
+      "loss": 0.7072,
+      "step": 168
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9719868173258005e-05,
+      "loss": 0.6327,
+      "step": 169
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.971751412429379e-05,
+      "loss": 0.5923,
+      "step": 170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.971516007532957e-05,
+      "loss": 0.5961,
+      "step": 171
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.971280602636535e-05,
+      "loss": 0.6435,
+      "step": 172
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9710451977401134e-05,
+      "loss": 0.6862,
+      "step": 173
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.970809792843691e-05,
+      "loss": 0.6389,
+      "step": 174
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9705743879472695e-05,
+      "loss": 0.5724,
+      "step": 175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.970338983050848e-05,
+      "loss": 0.6799,
+      "step": 176
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9701035781544256e-05,
+      "loss": 0.6455,
+      "step": 177
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.969868173258004e-05,
+      "loss": 0.6987,
+      "step": 178
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9696327683615824e-05,
+      "loss": 0.7139,
+      "step": 179
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.96939736346516e-05,
+      "loss": 0.6064,
+      "step": 180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9691619585687385e-05,
+      "loss": 0.6348,
+      "step": 181
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.968926553672317e-05,
+      "loss": 0.7095,
+      "step": 182
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9686911487758946e-05,
+      "loss": 0.6013,
+      "step": 183
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.968455743879473e-05,
+      "loss": 0.6665,
+      "step": 184
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9682203389830514e-05,
+      "loss": 0.6369,
+      "step": 185
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.967984934086629e-05,
+      "loss": 0.6243,
+      "step": 186
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9677495291902075e-05,
+      "loss": 0.6776,
+      "step": 187
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.967514124293786e-05,
+      "loss": 0.6432,
+      "step": 188
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9672787193973636e-05,
+      "loss": 0.635,
+      "step": 189
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.967043314500942e-05,
+      "loss": 0.6561,
+      "step": 190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.96680790960452e-05,
+      "loss": 0.6886,
+      "step": 191
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.966572504708098e-05,
+      "loss": 0.6533,
+      "step": 192
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9663370998116765e-05,
+      "loss": 0.6599,
+      "step": 193
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.966101694915254e-05,
+      "loss": 0.5998,
+      "step": 194
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9658662900188326e-05,
+      "loss": 0.6507,
+      "step": 195
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.965630885122411e-05,
+      "loss": 0.6565,
+      "step": 196
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.965395480225989e-05,
+      "loss": 0.6786,
+      "step": 197
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.965160075329567e-05,
+      "loss": 0.6513,
+      "step": 198
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9649246704331455e-05,
+      "loss": 0.6497,
+      "step": 199
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.964689265536723e-05,
+      "loss": 0.5744,
+      "step": 200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9644538606403016e-05,
+      "loss": 0.6643,
+      "step": 201
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.96421845574388e-05,
+      "loss": 0.6969,
+      "step": 202
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.963983050847458e-05,
+      "loss": 0.6324,
+      "step": 203
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.963747645951036e-05,
+      "loss": 0.6223,
+      "step": 204
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9635122410546145e-05,
+      "loss": 0.6021,
+      "step": 205
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.963276836158192e-05,
+      "loss": 0.6978,
+      "step": 206
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9630414312617706e-05,
+      "loss": 0.6619,
+      "step": 207
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.962806026365348e-05,
+      "loss": 0.6476,
+      "step": 208
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.962570621468927e-05,
+      "loss": 0.6361,
+      "step": 209
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.962335216572505e-05,
+      "loss": 0.6816,
+      "step": 210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.962099811676083e-05,
+      "loss": 0.6914,
+      "step": 211
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.961864406779661e-05,
+      "loss": 0.697,
+      "step": 212
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9616290018832396e-05,
+      "loss": 0.6442,
+      "step": 213
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.961393596986817e-05,
+      "loss": 0.603,
+      "step": 214
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.961158192090396e-05,
+      "loss": 0.6668,
+      "step": 215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.960922787193974e-05,
+      "loss": 0.6624,
+      "step": 216
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.960687382297552e-05,
+      "loss": 0.6418,
+      "step": 217
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.96045197740113e-05,
+      "loss": 0.6617,
+      "step": 218
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9602165725047086e-05,
+      "loss": 0.6046,
+      "step": 219
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.959981167608286e-05,
+      "loss": 0.6389,
+      "step": 220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.959745762711865e-05,
+      "loss": 0.6413,
+      "step": 221
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.959510357815443e-05,
+      "loss": 0.6584,
+      "step": 222
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.959274952919021e-05,
+      "loss": 0.6653,
+      "step": 223
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.959039548022599e-05,
+      "loss": 0.6143,
+      "step": 224
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9588041431261776e-05,
+      "loss": 0.641,
+      "step": 225
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.958568738229755e-05,
+      "loss": 0.6501,
+      "step": 226
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.958333333333334e-05,
+      "loss": 0.6718,
+      "step": 227
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9580979284369114e-05,
+      "loss": 0.6535,
+      "step": 228
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.95786252354049e-05,
+      "loss": 0.6758,
+      "step": 229
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.957627118644068e-05,
+      "loss": 0.6658,
+      "step": 230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.957391713747646e-05,
+      "loss": 0.6803,
+      "step": 231
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.957156308851224e-05,
+      "loss": 0.6456,
+      "step": 232
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.956920903954803e-05,
+      "loss": 0.6537,
+      "step": 233
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9566854990583804e-05,
+      "loss": 0.6759,
+      "step": 234
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.956450094161959e-05,
+      "loss": 0.6299,
+      "step": 235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.956214689265537e-05,
+      "loss": 0.6273,
+      "step": 236
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.955979284369115e-05,
+      "loss": 0.6789,
+      "step": 237
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.955743879472693e-05,
+      "loss": 0.6184,
+      "step": 238
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.955508474576272e-05,
+      "loss": 0.6668,
+      "step": 239
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9552730696798494e-05,
+      "loss": 0.5536,
+      "step": 240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.955037664783428e-05,
+      "loss": 0.5988,
+      "step": 241
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.954802259887006e-05,
+      "loss": 0.6694,
+      "step": 242
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.954566854990584e-05,
+      "loss": 0.7239,
+      "step": 243
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.954331450094162e-05,
+      "loss": 0.617,
+      "step": 244
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.95409604519774e-05,
+      "loss": 0.5939,
+      "step": 245
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9538606403013184e-05,
+      "loss": 0.6055,
+      "step": 246
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.953625235404897e-05,
+      "loss": 0.6361,
+      "step": 247
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9533898305084745e-05,
+      "loss": 0.6049,
+      "step": 248
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.953154425612053e-05,
+      "loss": 0.6858,
+      "step": 249
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.952919020715631e-05,
+      "loss": 0.6994,
+      "step": 250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.952683615819209e-05,
+      "loss": 0.6484,
+      "step": 251
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9524482109227874e-05,
+      "loss": 0.6407,
+      "step": 252
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.952212806026366e-05,
+      "loss": 0.6611,
+      "step": 253
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9519774011299435e-05,
+      "loss": 0.6912,
+      "step": 254
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.951741996233522e-05,
+      "loss": 0.6912,
+      "step": 255
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9515065913371e-05,
+      "loss": 0.6637,
+      "step": 256
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.951271186440678e-05,
+      "loss": 0.7289,
+      "step": 257
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9510357815442564e-05,
+      "loss": 0.6816,
+      "step": 258
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.950800376647835e-05,
+      "loss": 0.6343,
+      "step": 259
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9505649717514125e-05,
+      "loss": 0.6059,
+      "step": 260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.950329566854991e-05,
+      "loss": 0.6902,
+      "step": 261
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9500941619585686e-05,
+      "loss": 0.6675,
+      "step": 262
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.949858757062147e-05,
+      "loss": 0.647,
+      "step": 263
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9496233521657254e-05,
+      "loss": 0.6575,
+      "step": 264
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.949387947269303e-05,
+      "loss": 0.6798,
+      "step": 265
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9491525423728815e-05,
+      "loss": 0.636,
+      "step": 266
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.94891713747646e-05,
+      "loss": 0.6633,
+      "step": 267
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9486817325800376e-05,
+      "loss": 0.6894,
+      "step": 268
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.948446327683616e-05,
+      "loss": 0.6343,
+      "step": 269
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9482109227871944e-05,
+      "loss": 0.6325,
+      "step": 270
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.947975517890772e-05,
+      "loss": 0.6715,
+      "step": 271
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9477401129943505e-05,
+      "loss": 0.6327,
+      "step": 272
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.947504708097929e-05,
+      "loss": 0.6697,
+      "step": 273
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9472693032015066e-05,
+      "loss": 0.6723,
+      "step": 274
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.947033898305085e-05,
+      "loss": 0.6673,
+      "step": 275
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9467984934086634e-05,
+      "loss": 0.7047,
+      "step": 276
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.946563088512241e-05,
+      "loss": 0.6927,
+      "step": 277
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9463276836158195e-05,
+      "loss": 0.6481,
+      "step": 278
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.946092278719397e-05,
+      "loss": 0.6561,
+      "step": 279
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.945856873822976e-05,
+      "loss": 0.7036,
+      "step": 280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.945621468926554e-05,
+      "loss": 0.6105,
+      "step": 281
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.945386064030132e-05,
+      "loss": 0.7092,
+      "step": 282
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.94515065913371e-05,
+      "loss": 0.6625,
+      "step": 283
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9449152542372885e-05,
+      "loss": 0.6624,
+      "step": 284
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.944679849340866e-05,
+      "loss": 0.6214,
+      "step": 285
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9444444444444446e-05,
+      "loss": 0.6551,
+      "step": 286
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.944209039548023e-05,
+      "loss": 0.6794,
+      "step": 287
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.943973634651601e-05,
+      "loss": 0.7065,
+      "step": 288
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.943738229755179e-05,
+      "loss": 0.615,
+      "step": 289
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9435028248587575e-05,
+      "loss": 0.7009,
+      "step": 290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.943267419962335e-05,
+      "loss": 0.6497,
+      "step": 291
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9430320150659136e-05,
+      "loss": 0.6805,
+      "step": 292
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.942796610169492e-05,
+      "loss": 0.6581,
+      "step": 293
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.94256120527307e-05,
+      "loss": 0.6512,
+      "step": 294
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.942325800376648e-05,
+      "loss": 0.6609,
+      "step": 295
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.942090395480226e-05,
+      "loss": 0.6107,
+      "step": 296
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.941854990583805e-05,
+      "loss": 0.6746,
+      "step": 297
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9416195856873826e-05,
+      "loss": 0.6677,
+      "step": 298
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.94138418079096e-05,
+      "loss": 0.6349,
+      "step": 299
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9411487758945394e-05,
+      "loss": 0.6227,
+      "step": 300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.940913370998117e-05,
+      "loss": 0.6089,
+      "step": 301
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.940677966101695e-05,
+      "loss": 0.6874,
+      "step": 302
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.940442561205273e-05,
+      "loss": 0.6468,
+      "step": 303
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9402071563088516e-05,
+      "loss": 0.6929,
+      "step": 304
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.939971751412429e-05,
+      "loss": 0.6779,
+      "step": 305
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.939736346516008e-05,
+      "loss": 0.6786,
+      "step": 306
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.939500941619586e-05,
+      "loss": 0.592,
+      "step": 307
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.939265536723164e-05,
+      "loss": 0.6781,
+      "step": 308
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.939030131826742e-05,
+      "loss": 0.5925,
+      "step": 309
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9387947269303206e-05,
+      "loss": 0.6867,
+      "step": 310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.938559322033899e-05,
+      "loss": 0.6762,
+      "step": 311
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.938323917137477e-05,
+      "loss": 0.7195,
+      "step": 312
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9380885122410544e-05,
+      "loss": 0.5524,
+      "step": 313
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9378531073446335e-05,
+      "loss": 0.7096,
+      "step": 314
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.937617702448211e-05,
+      "loss": 0.6225,
+      "step": 315
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.937382297551789e-05,
+      "loss": 0.6398,
+      "step": 316
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.937146892655368e-05,
+      "loss": 0.6001,
+      "step": 317
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.936911487758946e-05,
+      "loss": 0.6783,
+      "step": 318
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9366760828625234e-05,
+      "loss": 0.6268,
+      "step": 319
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.936440677966102e-05,
+      "loss": 0.6048,
+      "step": 320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.93620527306968e-05,
+      "loss": 0.6385,
+      "step": 321
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.935969868173258e-05,
+      "loss": 0.6224,
+      "step": 322
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.935734463276836e-05,
+      "loss": 0.6818,
+      "step": 323
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.935499058380415e-05,
+      "loss": 0.6835,
+      "step": 324
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9352636534839924e-05,
+      "loss": 0.6548,
+      "step": 325
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.935028248587571e-05,
+      "loss": 0.6015,
+      "step": 326
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.934792843691149e-05,
+      "loss": 0.701,
+      "step": 327
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9345574387947276e-05,
+      "loss": 0.5609,
+      "step": 328
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.934322033898305e-05,
+      "loss": 0.5806,
+      "step": 329
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.934086629001883e-05,
+      "loss": 0.683,
+      "step": 330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.933851224105462e-05,
+      "loss": 0.5861,
+      "step": 331
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.93361581920904e-05,
+      "loss": 0.6397,
+      "step": 332
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9333804143126175e-05,
+      "loss": 0.6207,
+      "step": 333
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9331450094161966e-05,
+      "loss": 0.6493,
+      "step": 334
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.932909604519774e-05,
+      "loss": 0.6784,
+      "step": 335
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.932674199623352e-05,
+      "loss": 0.6625,
+      "step": 336
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9324387947269304e-05,
+      "loss": 0.6183,
+      "step": 337
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.932203389830509e-05,
+      "loss": 0.6148,
+      "step": 338
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9319679849340865e-05,
+      "loss": 0.659,
+      "step": 339
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.931732580037665e-05,
+      "loss": 0.669,
+      "step": 340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.931497175141243e-05,
+      "loss": 0.6191,
+      "step": 341
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.931261770244821e-05,
+      "loss": 0.7129,
+      "step": 342
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9310263653483994e-05,
+      "loss": 0.58,
+      "step": 343
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.930790960451978e-05,
+      "loss": 0.6315,
+      "step": 344
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.930555555555556e-05,
+      "loss": 0.687,
+      "step": 345
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.930320150659134e-05,
+      "loss": 0.6024,
+      "step": 346
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9300847457627116e-05,
+      "loss": 0.5917,
+      "step": 347
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.929849340866291e-05,
+      "loss": 0.6047,
+      "step": 348
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9296139359698684e-05,
+      "loss": 0.7005,
+      "step": 349
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.929378531073446e-05,
+      "loss": 0.6617,
+      "step": 350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.929143126177025e-05,
+      "loss": 0.6486,
+      "step": 351
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.928907721280603e-05,
+      "loss": 0.685,
+      "step": 352
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9286723163841806e-05,
+      "loss": 0.6286,
+      "step": 353
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.928436911487759e-05,
+      "loss": 0.645,
+      "step": 354
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9282015065913374e-05,
+      "loss": 0.6364,
+      "step": 355
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.927966101694915e-05,
+      "loss": 0.6552,
+      "step": 356
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9277306967984935e-05,
+      "loss": 0.6263,
+      "step": 357
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.927495291902072e-05,
+      "loss": 0.6486,
+      "step": 358
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.92725988700565e-05,
+      "loss": 0.6295,
+      "step": 359
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.927024482109228e-05,
+      "loss": 0.6131,
+      "step": 360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9267890772128064e-05,
+      "loss": 0.5999,
+      "step": 361
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.926553672316385e-05,
+      "loss": 0.6821,
+      "step": 362
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9263182674199625e-05,
+      "loss": 0.632,
+      "step": 363
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.92608286252354e-05,
+      "loss": 0.7303,
+      "step": 364
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.925847457627119e-05,
+      "loss": 0.6628,
+      "step": 365
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.925612052730697e-05,
+      "loss": 0.649,
+      "step": 366
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.925376647834275e-05,
+      "loss": 0.638,
+      "step": 367
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.925141242937854e-05,
+      "loss": 0.649,
+      "step": 368
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9249058380414315e-05,
+      "loss": 0.6283,
+      "step": 369
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.924670433145009e-05,
+      "loss": 0.6301,
+      "step": 370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9244350282485876e-05,
+      "loss": 0.6963,
+      "step": 371
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.924199623352166e-05,
+      "loss": 0.6623,
+      "step": 372
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.923964218455744e-05,
+      "loss": 0.6283,
+      "step": 373
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.923728813559322e-05,
+      "loss": 0.6544,
+      "step": 374
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9234934086629005e-05,
+      "loss": 0.6156,
+      "step": 375
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.923258003766479e-05,
+      "loss": 0.6522,
+      "step": 376
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9230225988700566e-05,
+      "loss": 0.6009,
+      "step": 377
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.922787193973635e-05,
+      "loss": 0.6749,
+      "step": 378
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9225517890772134e-05,
+      "loss": 0.6109,
+      "step": 379
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.922316384180791e-05,
+      "loss": 0.6607,
+      "step": 380
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.922080979284369e-05,
+      "loss": 0.6224,
+      "step": 381
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.921845574387948e-05,
+      "loss": 0.6427,
+      "step": 382
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9216101694915256e-05,
+      "loss": 0.6386,
+      "step": 383
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9213747645951034e-05,
+      "loss": 0.6477,
+      "step": 384
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9211393596986824e-05,
+      "loss": 0.5976,
+      "step": 385
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.92090395480226e-05,
+      "loss": 0.6395,
+      "step": 386
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.920668549905838e-05,
+      "loss": 0.5749,
+      "step": 387
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.920433145009416e-05,
+      "loss": 0.6893,
+      "step": 388
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9201977401129946e-05,
+      "loss": 0.6841,
+      "step": 389
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.919962335216573e-05,
+      "loss": 0.664,
+      "step": 390
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.919726930320151e-05,
+      "loss": 0.6352,
+      "step": 391
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.919491525423729e-05,
+      "loss": 0.6586,
+      "step": 392
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9192561205273075e-05,
+      "loss": 0.6588,
+      "step": 393
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.919020715630885e-05,
+      "loss": 0.6334,
+      "step": 394
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9187853107344636e-05,
+      "loss": 0.6313,
+      "step": 395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.918549905838042e-05,
+      "loss": 0.6282,
+      "step": 396
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.91831450094162e-05,
+      "loss": 0.5805,
+      "step": 397
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9180790960451975e-05,
+      "loss": 0.6239,
+      "step": 398
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9178436911487765e-05,
+      "loss": 0.6273,
+      "step": 399
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.917608286252354e-05,
+      "loss": 0.6454,
+      "step": 400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.917372881355932e-05,
+      "loss": 0.6149,
+      "step": 401
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.917137476459511e-05,
+      "loss": 0.5899,
+      "step": 402
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.916902071563089e-05,
+      "loss": 0.5464,
+      "step": 403
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9166666666666665e-05,
+      "loss": 0.7368,
+      "step": 404
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.916431261770245e-05,
+      "loss": 0.6547,
+      "step": 405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.916195856873823e-05,
+      "loss": 0.6531,
+      "step": 406
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9159604519774016e-05,
+      "loss": 0.6269,
+      "step": 407
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9157250470809794e-05,
+      "loss": 0.6672,
+      "step": 408
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.915489642184558e-05,
+      "loss": 0.5971,
+      "step": 409
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.915254237288136e-05,
+      "loss": 0.6463,
+      "step": 410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.915018832391714e-05,
+      "loss": 0.7032,
+      "step": 411
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.914783427495292e-05,
+      "loss": 0.6895,
+      "step": 412
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9145480225988706e-05,
+      "loss": 0.6581,
+      "step": 413
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9143126177024484e-05,
+      "loss": 0.6225,
+      "step": 414
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.914077212806026e-05,
+      "loss": 0.7009,
+      "step": 415
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.913841807909605e-05,
+      "loss": 0.5491,
+      "step": 416
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.913606403013183e-05,
+      "loss": 0.6789,
+      "step": 417
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9133709981167606e-05,
+      "loss": 0.6756,
+      "step": 418
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9131355932203396e-05,
+      "loss": 0.7268,
+      "step": 419
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9129001883239174e-05,
+      "loss": 0.6167,
+      "step": 420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.912664783427495e-05,
+      "loss": 0.602,
+      "step": 421
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9124293785310735e-05,
+      "loss": 0.6567,
+      "step": 422
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.912193973634652e-05,
+      "loss": 0.6435,
+      "step": 423
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.91195856873823e-05,
+      "loss": 0.6507,
+      "step": 424
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.911723163841808e-05,
+      "loss": 0.6893,
+      "step": 425
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9114877589453864e-05,
+      "loss": 0.6929,
+      "step": 426
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.911252354048965e-05,
+      "loss": 0.6955,
+      "step": 427
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9110169491525425e-05,
+      "loss": 0.6161,
+      "step": 428
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.910781544256121e-05,
+      "loss": 0.5958,
+      "step": 429
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.910546139359699e-05,
+      "loss": 0.6269,
+      "step": 430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.910310734463277e-05,
+      "loss": 0.6261,
+      "step": 431
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.910075329566855e-05,
+      "loss": 0.6365,
+      "step": 432
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.909839924670434e-05,
+      "loss": 0.6631,
+      "step": 433
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9096045197740115e-05,
+      "loss": 0.6346,
+      "step": 434
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.909369114877589e-05,
+      "loss": 0.698,
+      "step": 435
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.909133709981168e-05,
+      "loss": 0.6358,
+      "step": 436
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.908898305084746e-05,
+      "loss": 0.6212,
+      "step": 437
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9086629001883244e-05,
+      "loss": 0.71,
+      "step": 438
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.908427495291902e-05,
+      "loss": 0.7036,
+      "step": 439
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9081920903954805e-05,
+      "loss": 0.7036,
+      "step": 440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.907956685499059e-05,
+      "loss": 0.6116,
+      "step": 441
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9077212806026366e-05,
+      "loss": 0.6079,
+      "step": 442
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.907485875706215e-05,
+      "loss": 0.5471,
+      "step": 443
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9072504708097934e-05,
+      "loss": 0.647,
+      "step": 444
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.907015065913371e-05,
+      "loss": 0.6526,
+      "step": 445
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9067796610169495e-05,
+      "loss": 0.6373,
+      "step": 446
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.906544256120528e-05,
+      "loss": 0.6455,
+      "step": 447
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9063088512241056e-05,
+      "loss": 0.6393,
+      "step": 448
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.906073446327684e-05,
+      "loss": 0.6509,
+      "step": 449
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9058380414312623e-05,
+      "loss": 0.6479,
+      "step": 450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.90560263653484e-05,
+      "loss": 0.653,
+      "step": 451
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.905367231638418e-05,
+      "loss": 0.6449,
+      "step": 452
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.905131826741997e-05,
+      "loss": 0.6267,
+      "step": 453
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9048964218455746e-05,
+      "loss": 0.621,
+      "step": 454
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.904661016949153e-05,
+      "loss": 0.5811,
+      "step": 455
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.904425612052731e-05,
+      "loss": 0.5734,
+      "step": 456
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.904190207156309e-05,
+      "loss": 0.65,
+      "step": 457
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9039548022598875e-05,
+      "loss": 0.6375,
+      "step": 458
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.903719397363465e-05,
+      "loss": 0.5783,
+      "step": 459
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9034839924670436e-05,
+      "loss": 0.614,
+      "step": 460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.903248587570622e-05,
+      "loss": 0.6243,
+      "step": 461
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9030131826742e-05,
+      "loss": 0.6038,
+      "step": 462
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.902777777777778e-05,
+      "loss": 0.6104,
+      "step": 463
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9025423728813565e-05,
+      "loss": 0.6617,
+      "step": 464
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.902306967984934e-05,
+      "loss": 0.6496,
+      "step": 465
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9020715630885126e-05,
+      "loss": 0.64,
+      "step": 466
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.901836158192091e-05,
+      "loss": 0.6478,
+      "step": 467
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.901600753295669e-05,
+      "loss": 0.603,
+      "step": 468
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.901365348399247e-05,
+      "loss": 0.6536,
+      "step": 469
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9011299435028255e-05,
+      "loss": 0.6057,
+      "step": 470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.900894538606403e-05,
+      "loss": 0.6066,
+      "step": 471
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9006591337099816e-05,
+      "loss": 0.6752,
+      "step": 472
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.900423728813559e-05,
+      "loss": 0.6047,
+      "step": 473
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.900188323917138e-05,
+      "loss": 0.621,
+      "step": 474
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.899952919020716e-05,
+      "loss": 0.6663,
+      "step": 475
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.899717514124294e-05,
+      "loss": 0.6173,
+      "step": 476
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.899482109227872e-05,
+      "loss": 0.6041,
+      "step": 477
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8992467043314506e-05,
+      "loss": 0.6176,
+      "step": 478
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.899011299435028e-05,
+      "loss": 0.7024,
+      "step": 479
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.898775894538607e-05,
+      "loss": 0.679,
+      "step": 480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.898540489642185e-05,
+      "loss": 0.6845,
+      "step": 481
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.898305084745763e-05,
+      "loss": 0.6587,
+      "step": 482
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.898069679849341e-05,
+      "loss": 0.6679,
+      "step": 483
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8978342749529196e-05,
+      "loss": 0.6254,
+      "step": 484
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.897598870056497e-05,
+      "loss": 0.6678,
+      "step": 485
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.897363465160076e-05,
+      "loss": 0.6302,
+      "step": 486
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.897128060263654e-05,
+      "loss": 0.653,
+      "step": 487
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.896892655367232e-05,
+      "loss": 0.6054,
+      "step": 488
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.89665725047081e-05,
+      "loss": 0.581,
+      "step": 489
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.896421845574388e-05,
+      "loss": 0.6714,
+      "step": 490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.896186440677966e-05,
+      "loss": 0.608,
+      "step": 491
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.895951035781545e-05,
+      "loss": 0.6479,
+      "step": 492
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8957156308851224e-05,
+      "loss": 0.6083,
+      "step": 493
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.895480225988701e-05,
+      "loss": 0.6047,
+      "step": 494
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.895244821092279e-05,
+      "loss": 0.7008,
+      "step": 495
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.895009416195857e-05,
+      "loss": 0.657,
+      "step": 496
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.894774011299435e-05,
+      "loss": 0.6485,
+      "step": 497
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.894538606403014e-05,
+      "loss": 0.5921,
+      "step": 498
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8943032015065914e-05,
+      "loss": 0.6264,
+      "step": 499
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.89406779661017e-05,
+      "loss": 0.6239,
+      "step": 500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.893832391713748e-05,
+      "loss": 0.5823,
+      "step": 501
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.893596986817326e-05,
+      "loss": 0.6853,
+      "step": 502
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.893361581920904e-05,
+      "loss": 0.5856,
+      "step": 503
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.893126177024483e-05,
+      "loss": 0.6548,
+      "step": 504
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8928907721280604e-05,
+      "loss": 0.5776,
+      "step": 505
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.892655367231639e-05,
+      "loss": 0.6032,
+      "step": 506
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8924199623352165e-05,
+      "loss": 0.6366,
+      "step": 507
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.892184557438795e-05,
+      "loss": 0.6663,
+      "step": 508
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.891949152542373e-05,
+      "loss": 0.6061,
+      "step": 509
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.891713747645951e-05,
+      "loss": 0.6634,
+      "step": 510
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8914783427495294e-05,
+      "loss": 0.6706,
+      "step": 511
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.891242937853108e-05,
+      "loss": 0.6159,
+      "step": 512
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8910075329566855e-05,
+      "loss": 0.7242,
+      "step": 513
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.890772128060264e-05,
+      "loss": 0.6019,
+      "step": 514
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.890536723163842e-05,
+      "loss": 0.6863,
+      "step": 515
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.89030131826742e-05,
+      "loss": 0.6445,
+      "step": 516
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8900659133709984e-05,
+      "loss": 0.6969,
+      "step": 517
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.889830508474577e-05,
+      "loss": 0.6388,
+      "step": 518
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8895951035781545e-05,
+      "loss": 0.6494,
+      "step": 519
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.889359698681733e-05,
+      "loss": 0.6212,
+      "step": 520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.889124293785311e-05,
+      "loss": 0.62,
+      "step": 521
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.888888888888889e-05,
+      "loss": 0.6494,
+      "step": 522
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8886534839924674e-05,
+      "loss": 0.6475,
+      "step": 523
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.888418079096045e-05,
+      "loss": 0.5801,
+      "step": 524
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8881826741996235e-05,
+      "loss": 0.6253,
+      "step": 525
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.887947269303202e-05,
+      "loss": 0.6146,
+      "step": 526
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8877118644067796e-05,
+      "loss": 0.6237,
+      "step": 527
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.887476459510358e-05,
+      "loss": 0.5503,
+      "step": 528
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8872410546139364e-05,
+      "loss": 0.6442,
+      "step": 529
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.887005649717514e-05,
+      "loss": 0.6076,
+      "step": 530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8867702448210925e-05,
+      "loss": 0.6669,
+      "step": 531
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.886534839924671e-05,
+      "loss": 0.6181,
+      "step": 532
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8862994350282486e-05,
+      "loss": 0.7584,
+      "step": 533
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.886064030131827e-05,
+      "loss": 0.718,
+      "step": 534
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8858286252354054e-05,
+      "loss": 0.6001,
+      "step": 535
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.885593220338983e-05,
+      "loss": 0.5508,
+      "step": 536
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8853578154425615e-05,
+      "loss": 0.5833,
+      "step": 537
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.88512241054614e-05,
+      "loss": 0.5685,
+      "step": 538
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8848870056497176e-05,
+      "loss": 0.5666,
+      "step": 539
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.884651600753296e-05,
+      "loss": 0.6498,
+      "step": 540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8844161958568744e-05,
+      "loss": 0.6808,
+      "step": 541
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.884180790960452e-05,
+      "loss": 0.5837,
+      "step": 542
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8839453860640305e-05,
+      "loss": 0.6119,
+      "step": 543
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.883709981167608e-05,
+      "loss": 0.5911,
+      "step": 544
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8834745762711866e-05,
+      "loss": 0.6456,
+      "step": 545
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.883239171374765e-05,
+      "loss": 0.6144,
+      "step": 546
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.883003766478343e-05,
+      "loss": 0.6007,
+      "step": 547
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.882768361581921e-05,
+      "loss": 0.6769,
+      "step": 548
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8825329566854995e-05,
+      "loss": 0.6289,
+      "step": 549
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.882297551789077e-05,
+      "loss": 0.5484,
+      "step": 550
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8820621468926556e-05,
+      "loss": 0.6282,
+      "step": 551
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.881826741996234e-05,
+      "loss": 0.6054,
+      "step": 552
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.881591337099812e-05,
+      "loss": 0.5775,
+      "step": 553
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.88135593220339e-05,
+      "loss": 0.5558,
+      "step": 554
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8811205273069685e-05,
+      "loss": 0.5763,
+      "step": 555
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.880885122410546e-05,
+      "loss": 0.6572,
+      "step": 556
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8806497175141246e-05,
+      "loss": 0.5574,
+      "step": 557
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.880414312617703e-05,
+      "loss": 0.6263,
+      "step": 558
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.880178907721281e-05,
+      "loss": 0.5932,
+      "step": 559
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.879943502824859e-05,
+      "loss": 0.6124,
+      "step": 560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.879708097928437e-05,
+      "loss": 0.6439,
+      "step": 561
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.879472693032015e-05,
+      "loss": 0.6052,
+      "step": 562
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8792372881355936e-05,
+      "loss": 0.6574,
+      "step": 563
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.879001883239171e-05,
+      "loss": 0.6491,
+      "step": 564
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.87876647834275e-05,
+      "loss": 0.6192,
+      "step": 565
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.878531073446328e-05,
+      "loss": 0.7114,
+      "step": 566
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.878295668549906e-05,
+      "loss": 0.6166,
+      "step": 567
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.878060263653484e-05,
+      "loss": 0.6218,
+      "step": 568
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8778248587570626e-05,
+      "loss": 0.6508,
+      "step": 569
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.87758945386064e-05,
+      "loss": 0.6095,
+      "step": 570
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.877354048964219e-05,
+      "loss": 0.6798,
+      "step": 571
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.877118644067797e-05,
+      "loss": 0.6171,
+      "step": 572
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.876883239171375e-05,
+      "loss": 0.6273,
+      "step": 573
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.876647834274953e-05,
+      "loss": 0.5677,
+      "step": 574
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8764124293785316e-05,
+      "loss": 0.6507,
+      "step": 575
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.876177024482109e-05,
+      "loss": 0.6138,
+      "step": 576
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.875941619585688e-05,
+      "loss": 0.6194,
+      "step": 577
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8757062146892654e-05,
+      "loss": 0.6541,
+      "step": 578
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.875470809792844e-05,
+      "loss": 0.6229,
+      "step": 579
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.875235404896422e-05,
+      "loss": 0.6128,
+      "step": 580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.875e-05,
+      "loss": 0.6406,
+      "step": 581
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.874764595103578e-05,
+      "loss": 0.617,
+      "step": 582
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.874529190207157e-05,
+      "loss": 0.6187,
+      "step": 583
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8742937853107344e-05,
+      "loss": 0.5935,
+      "step": 584
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.874058380414313e-05,
+      "loss": 0.5446,
+      "step": 585
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.873822975517891e-05,
+      "loss": 0.604,
+      "step": 586
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.873587570621469e-05,
+      "loss": 0.6234,
+      "step": 587
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.873352165725047e-05,
+      "loss": 0.668,
+      "step": 588
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.873116760828626e-05,
+      "loss": 0.6124,
+      "step": 589
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8728813559322034e-05,
+      "loss": 0.7136,
+      "step": 590
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.872645951035782e-05,
+      "loss": 0.6202,
+      "step": 591
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.87241054613936e-05,
+      "loss": 0.6154,
+      "step": 592
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.872175141242938e-05,
+      "loss": 0.6142,
+      "step": 593
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.871939736346516e-05,
+      "loss": 0.6041,
+      "step": 594
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.871704331450094e-05,
+      "loss": 0.6913,
+      "step": 595
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.871468926553673e-05,
+      "loss": 0.7386,
+      "step": 596
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.871233521657251e-05,
+      "loss": 0.6175,
+      "step": 597
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8709981167608285e-05,
+      "loss": 0.6201,
+      "step": 598
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.870762711864407e-05,
+      "loss": 0.6651,
+      "step": 599
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.870527306967985e-05,
+      "loss": 0.5681,
+      "step": 600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.870291902071563e-05,
+      "loss": 0.5604,
+      "step": 601
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8700564971751414e-05,
+      "loss": 0.5787,
+      "step": 602
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.86982109227872e-05,
+      "loss": 0.6467,
+      "step": 603
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8695856873822975e-05,
+      "loss": 0.6555,
+      "step": 604
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.869350282485876e-05,
+      "loss": 0.6622,
+      "step": 605
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.869114877589454e-05,
+      "loss": 0.6308,
+      "step": 606
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.868879472693032e-05,
+      "loss": 0.5625,
+      "step": 607
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8686440677966104e-05,
+      "loss": 0.5904,
+      "step": 608
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.868408662900189e-05,
+      "loss": 0.5745,
+      "step": 609
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8681732580037665e-05,
+      "loss": 0.5678,
+      "step": 610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.867937853107345e-05,
+      "loss": 0.6277,
+      "step": 611
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8677024482109226e-05,
+      "loss": 0.6446,
+      "step": 612
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.867467043314502e-05,
+      "loss": 0.6801,
+      "step": 613
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8672316384180794e-05,
+      "loss": 0.5571,
+      "step": 614
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.866996233521657e-05,
+      "loss": 0.638,
+      "step": 615
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.866760828625236e-05,
+      "loss": 0.6263,
+      "step": 616
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.866525423728814e-05,
+      "loss": 0.5882,
+      "step": 617
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8662900188323916e-05,
+      "loss": 0.5585,
+      "step": 618
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.86605461393597e-05,
+      "loss": 0.6278,
+      "step": 619
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8658192090395484e-05,
+      "loss": 0.6429,
+      "step": 620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.865583804143126e-05,
+      "loss": 0.5504,
+      "step": 621
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8653483992467045e-05,
+      "loss": 0.6946,
+      "step": 622
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.865112994350283e-05,
+      "loss": 0.6207,
+      "step": 623
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8648775894538606e-05,
+      "loss": 0.5106,
+      "step": 624
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.864642184557439e-05,
+      "loss": 0.5856,
+      "step": 625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8644067796610174e-05,
+      "loss": 0.6857,
+      "step": 626
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.864171374764596e-05,
+      "loss": 0.6064,
+      "step": 627
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8639359698681735e-05,
+      "loss": 0.638,
+      "step": 628
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.863700564971751e-05,
+      "loss": 0.6148,
+      "step": 629
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.86346516007533e-05,
+      "loss": 0.6616,
+      "step": 630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.863229755178908e-05,
+      "loss": 0.5755,
+      "step": 631
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.862994350282486e-05,
+      "loss": 0.6428,
+      "step": 632
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.862758945386065e-05,
+      "loss": 0.6739,
+      "step": 633
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8625235404896425e-05,
+      "loss": 0.5954,
+      "step": 634
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.86228813559322e-05,
+      "loss": 0.6145,
+      "step": 635
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8620527306967986e-05,
+      "loss": 0.5675,
+      "step": 636
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.861817325800377e-05,
+      "loss": 0.6535,
+      "step": 637
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.861581920903955e-05,
+      "loss": 0.6434,
+      "step": 638
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.861346516007533e-05,
+      "loss": 0.6126,
+      "step": 639
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8611111111111115e-05,
+      "loss": 0.6132,
+      "step": 640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.860875706214689e-05,
+      "loss": 0.5778,
+      "step": 641
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8606403013182676e-05,
+      "loss": 0.5499,
+      "step": 642
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.860404896421846e-05,
+      "loss": 0.6425,
+      "step": 643
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8601694915254244e-05,
+      "loss": 0.7085,
+      "step": 644
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.859934086629002e-05,
+      "loss": 0.5235,
+      "step": 645
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.85969868173258e-05,
+      "loss": 0.6908,
+      "step": 646
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.859463276836159e-05,
+      "loss": 0.6036,
+      "step": 647
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8592278719397366e-05,
+      "loss": 0.6394,
+      "step": 648
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.858992467043314e-05,
+      "loss": 0.7065,
+      "step": 649
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8587570621468934e-05,
+      "loss": 0.6445,
+      "step": 650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.858521657250471e-05,
+      "loss": 0.6741,
+      "step": 651
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.858286252354049e-05,
+      "loss": 0.6161,
+      "step": 652
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.858050847457627e-05,
+      "loss": 0.6138,
+      "step": 653
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8578154425612056e-05,
+      "loss": 0.7495,
+      "step": 654
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.857580037664783e-05,
+      "loss": 0.5824,
+      "step": 655
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.857344632768362e-05,
+      "loss": 0.6073,
+      "step": 656
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.85710922787194e-05,
+      "loss": 0.6253,
+      "step": 657
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.856873822975518e-05,
+      "loss": 0.5975,
+      "step": 658
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.856638418079096e-05,
+      "loss": 0.6669,
+      "step": 659
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8564030131826746e-05,
+      "loss": 0.543,
+      "step": 660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.856167608286253e-05,
+      "loss": 0.6545,
+      "step": 661
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.855932203389831e-05,
+      "loss": 0.5986,
+      "step": 662
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8556967984934084e-05,
+      "loss": 0.576,
+      "step": 663
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8554613935969875e-05,
+      "loss": 0.6801,
+      "step": 664
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.855225988700565e-05,
+      "loss": 0.5862,
+      "step": 665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.854990583804143e-05,
+      "loss": 0.5303,
+      "step": 666
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.854755178907722e-05,
+      "loss": 0.6014,
+      "step": 667
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8545197740113e-05,
+      "loss": 0.678,
+      "step": 668
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8542843691148774e-05,
+      "loss": 0.7181,
+      "step": 669
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.854048964218456e-05,
+      "loss": 0.6667,
+      "step": 670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.853813559322034e-05,
+      "loss": 0.6387,
+      "step": 671
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.853578154425612e-05,
+      "loss": 0.6104,
+      "step": 672
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.85334274952919e-05,
+      "loss": 0.5413,
+      "step": 673
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.853107344632769e-05,
+      "loss": 0.6183,
+      "step": 674
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.852871939736347e-05,
+      "loss": 0.574,
+      "step": 675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.852636534839925e-05,
+      "loss": 0.6834,
+      "step": 676
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.852401129943503e-05,
+      "loss": 0.6254,
+      "step": 677
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8521657250470816e-05,
+      "loss": 0.5562,
+      "step": 678
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.851930320150659e-05,
+      "loss": 0.5903,
+      "step": 679
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.851694915254237e-05,
+      "loss": 0.5961,
+      "step": 680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.851459510357816e-05,
+      "loss": 0.6249,
+      "step": 681
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.851224105461394e-05,
+      "loss": 0.6372,
+      "step": 682
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8509887005649715e-05,
+      "loss": 0.5141,
+      "step": 683
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8507532956685506e-05,
+      "loss": 0.6379,
+      "step": 684
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.850517890772128e-05,
+      "loss": 0.6042,
+      "step": 685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.850282485875706e-05,
+      "loss": 0.6922,
+      "step": 686
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8500470809792844e-05,
+      "loss": 0.6356,
+      "step": 687
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.849811676082863e-05,
+      "loss": 0.6297,
+      "step": 688
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8495762711864405e-05,
+      "loss": 0.6763,
+      "step": 689
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.849340866290019e-05,
+      "loss": 0.6197,
+      "step": 690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.849105461393597e-05,
+      "loss": 0.6232,
+      "step": 691
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.848870056497176e-05,
+      "loss": 0.6219,
+      "step": 692
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8486346516007534e-05,
+      "loss": 0.6014,
+      "step": 693
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.848399246704332e-05,
+      "loss": 0.6412,
+      "step": 694
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.84816384180791e-05,
+      "loss": 0.5831,
+      "step": 695
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.847928436911488e-05,
+      "loss": 0.5874,
+      "step": 696
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8476930320150657e-05,
+      "loss": 0.6024,
+      "step": 697
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.847457627118645e-05,
+      "loss": 0.727,
+      "step": 698
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8472222222222224e-05,
+      "loss": 0.6361,
+      "step": 699
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8469868173258e-05,
+      "loss": 0.6107,
+      "step": 700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.846751412429379e-05,
+      "loss": 0.6151,
+      "step": 701
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.846516007532957e-05,
+      "loss": 0.5532,
+      "step": 702
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8462806026365347e-05,
+      "loss": 0.6207,
+      "step": 703
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.846045197740113e-05,
+      "loss": 0.6307,
+      "step": 704
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8458097928436914e-05,
+      "loss": 0.6683,
+      "step": 705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.84557438794727e-05,
+      "loss": 0.6133,
+      "step": 706
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8453389830508475e-05,
+      "loss": 0.5911,
+      "step": 707
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.845103578154426e-05,
+      "loss": 0.5459,
+      "step": 708
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.844868173258004e-05,
+      "loss": 0.6459,
+      "step": 709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.844632768361582e-05,
+      "loss": 0.6029,
+      "step": 710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8443973634651604e-05,
+      "loss": 0.6229,
+      "step": 711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.844161958568739e-05,
+      "loss": 0.6038,
+      "step": 712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8439265536723165e-05,
+      "loss": 0.6234,
+      "step": 713
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.843691148775894e-05,
+      "loss": 0.5612,
+      "step": 714
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.843455743879473e-05,
+      "loss": 0.6934,
+      "step": 715
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.843220338983051e-05,
+      "loss": 0.7028,
+      "step": 716
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.842984934086629e-05,
+      "loss": 0.685,
+      "step": 717
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.842749529190208e-05,
+      "loss": 0.6815,
+      "step": 718
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8425141242937855e-05,
+      "loss": 0.6432,
+      "step": 719
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.842278719397363e-05,
+      "loss": 0.6114,
+      "step": 720
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8420433145009417e-05,
+      "loss": 0.7284,
+      "step": 721
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.84180790960452e-05,
+      "loss": 0.5731,
+      "step": 722
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8415725047080984e-05,
+      "loss": 0.6295,
+      "step": 723
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.841337099811676e-05,
+      "loss": 0.5874,
+      "step": 724
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8411016949152545e-05,
+      "loss": 0.7564,
+      "step": 725
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.840866290018833e-05,
+      "loss": 0.6438,
+      "step": 726
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8406308851224106e-05,
+      "loss": 0.6027,
+      "step": 727
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.840395480225989e-05,
+      "loss": 0.6746,
+      "step": 728
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8401600753295674e-05,
+      "loss": 0.661,
+      "step": 729
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.839924670433145e-05,
+      "loss": 0.5698,
+      "step": 730
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.839689265536723e-05,
+      "loss": 0.6258,
+      "step": 731
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.839453860640302e-05,
+      "loss": 0.6249,
+      "step": 732
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8392184557438796e-05,
+      "loss": 0.6278,
+      "step": 733
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8389830508474574e-05,
+      "loss": 0.5305,
+      "step": 734
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8387476459510364e-05,
+      "loss": 0.5569,
+      "step": 735
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.838512241054614e-05,
+      "loss": 0.6538,
+      "step": 736
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.838276836158192e-05,
+      "loss": 0.5709,
+      "step": 737
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.83804143126177e-05,
+      "loss": 0.5671,
+      "step": 738
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8378060263653486e-05,
+      "loss": 0.6532,
+      "step": 739
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.837570621468927e-05,
+      "loss": 0.6429,
+      "step": 740
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.837335216572505e-05,
+      "loss": 0.5366,
+      "step": 741
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.837099811676083e-05,
+      "loss": 0.6145,
+      "step": 742
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8368644067796615e-05,
+      "loss": 0.6061,
+      "step": 743
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.836629001883239e-05,
+      "loss": 0.7353,
+      "step": 744
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8363935969868176e-05,
+      "loss": 0.6444,
+      "step": 745
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.836158192090396e-05,
+      "loss": 0.5201,
+      "step": 746
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.835922787193974e-05,
+      "loss": 0.6515,
+      "step": 747
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8356873822975515e-05,
+      "loss": 0.698,
+      "step": 748
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8354519774011305e-05,
+      "loss": 0.5823,
+      "step": 749
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.835216572504708e-05,
+      "loss": 0.5327,
+      "step": 750
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.834981167608286e-05,
+      "loss": 0.6294,
+      "step": 751
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.834745762711865e-05,
+      "loss": 0.5886,
+      "step": 752
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.834510357815443e-05,
+      "loss": 0.5674,
+      "step": 753
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.834274952919021e-05,
+      "loss": 0.593,
+      "step": 754
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.834039548022599e-05,
+      "loss": 0.6456,
+      "step": 755
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.833804143126177e-05,
+      "loss": 0.6764,
+      "step": 756
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8335687382297556e-05,
+      "loss": 0.6694,
+      "step": 757
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8333333333333334e-05,
+      "loss": 0.5505,
+      "step": 758
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.833097928436912e-05,
+      "loss": 0.6387,
+      "step": 759
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.83286252354049e-05,
+      "loss": 0.6097,
+      "step": 760
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.832627118644068e-05,
+      "loss": 0.5811,
+      "step": 761
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.832391713747646e-05,
+      "loss": 0.617,
+      "step": 762
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8321563088512246e-05,
+      "loss": 0.5836,
+      "step": 763
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8319209039548024e-05,
+      "loss": 0.6217,
+      "step": 764
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.831685499058381e-05,
+      "loss": 0.5317,
+      "step": 765
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.831450094161959e-05,
+      "loss": 0.5566,
+      "step": 766
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.831214689265537e-05,
+      "loss": 0.5916,
+      "step": 767
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8309792843691146e-05,
+      "loss": 0.6032,
+      "step": 768
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8307438794726936e-05,
+      "loss": 0.5431,
+      "step": 769
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8305084745762714e-05,
+      "loss": 0.564,
+      "step": 770
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.83027306967985e-05,
+      "loss": 0.5869,
+      "step": 771
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8300376647834275e-05,
+      "loss": 0.6008,
+      "step": 772
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.829802259887006e-05,
+      "loss": 0.5735,
+      "step": 773
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.829566854990584e-05,
+      "loss": 0.6322,
+      "step": 774
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.829331450094162e-05,
+      "loss": 0.577,
+      "step": 775
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8290960451977404e-05,
+      "loss": 0.5537,
+      "step": 776
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.828860640301319e-05,
+      "loss": 0.64,
+      "step": 777
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8286252354048965e-05,
+      "loss": 0.7971,
+      "step": 778
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.828389830508475e-05,
+      "loss": 0.739,
+      "step": 779
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.828154425612053e-05,
+      "loss": 0.5542,
+      "step": 780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.827919020715631e-05,
+      "loss": 0.5621,
+      "step": 781
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8276836158192094e-05,
+      "loss": 0.6847,
+      "step": 782
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.827448210922788e-05,
+      "loss": 0.618,
+      "step": 783
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8272128060263655e-05,
+      "loss": 0.559,
+      "step": 784
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.826977401129944e-05,
+      "loss": 0.5478,
+      "step": 785
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.826741996233522e-05,
+      "loss": 0.5501,
+      "step": 786
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8265065913371e-05,
+      "loss": 0.647,
+      "step": 787
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8262711864406784e-05,
+      "loss": 0.6673,
+      "step": 788
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.826035781544256e-05,
+      "loss": 0.5661,
+      "step": 789
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8258003766478345e-05,
+      "loss": 0.6418,
+      "step": 790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.825564971751413e-05,
+      "loss": 0.5894,
+      "step": 791
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8253295668549906e-05,
+      "loss": 0.6125,
+      "step": 792
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.825094161958569e-05,
+      "loss": 0.6207,
+      "step": 793
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8248587570621474e-05,
+      "loss": 0.6351,
+      "step": 794
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.824623352165725e-05,
+      "loss": 0.612,
+      "step": 795
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8243879472693035e-05,
+      "loss": 0.6421,
+      "step": 796
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.824152542372882e-05,
+      "loss": 0.5526,
+      "step": 797
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8239171374764596e-05,
+      "loss": 0.6595,
+      "step": 798
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.823681732580038e-05,
+      "loss": 0.7017,
+      "step": 799
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8234463276836164e-05,
+      "loss": 0.6451,
+      "step": 800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.823210922787194e-05,
+      "loss": 0.618,
+      "step": 801
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8229755178907725e-05,
+      "loss": 0.5809,
+      "step": 802
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.822740112994351e-05,
+      "loss": 0.6161,
+      "step": 803
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8225047080979286e-05,
+      "loss": 0.6176,
+      "step": 804
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.822269303201507e-05,
+      "loss": 0.6457,
+      "step": 805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.822033898305085e-05,
+      "loss": 0.6448,
+      "step": 806
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.821798493408663e-05,
+      "loss": 0.6331,
+      "step": 807
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8215630885122415e-05,
+      "loss": 0.645,
+      "step": 808
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.821327683615819e-05,
+      "loss": 0.5629,
+      "step": 809
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8210922787193976e-05,
+      "loss": 0.5951,
+      "step": 810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.820856873822976e-05,
+      "loss": 0.6019,
+      "step": 811
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.820621468926554e-05,
+      "loss": 0.6395,
+      "step": 812
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.820386064030132e-05,
+      "loss": 0.5849,
+      "step": 813
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8201506591337105e-05,
+      "loss": 0.5712,
+      "step": 814
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.819915254237288e-05,
+      "loss": 0.5639,
+      "step": 815
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8196798493408666e-05,
+      "loss": 0.4692,
+      "step": 816
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.819444444444445e-05,
+      "loss": 0.6298,
+      "step": 817
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.819209039548023e-05,
+      "loss": 0.6205,
+      "step": 818
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.818973634651601e-05,
+      "loss": 0.6453,
+      "step": 819
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8187382297551795e-05,
+      "loss": 0.6607,
+      "step": 820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.818502824858757e-05,
+      "loss": 0.6248,
+      "step": 821
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8182674199623356e-05,
+      "loss": 0.5185,
+      "step": 822
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.818032015065913e-05,
+      "loss": 0.5788,
+      "step": 823
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.817796610169492e-05,
+      "loss": 0.5363,
+      "step": 824
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.81756120527307e-05,
+      "loss": 0.5532,
+      "step": 825
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.817325800376648e-05,
+      "loss": 0.5867,
+      "step": 826
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.817090395480226e-05,
+      "loss": 0.629,
+      "step": 827
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8168549905838046e-05,
+      "loss": 0.5829,
+      "step": 828
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.816619585687382e-05,
+      "loss": 0.6395,
+      "step": 829
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.816384180790961e-05,
+      "loss": 0.7104,
+      "step": 830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.816148775894539e-05,
+      "loss": 0.6839,
+      "step": 831
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.815913370998117e-05,
+      "loss": 0.5375,
+      "step": 832
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.815677966101695e-05,
+      "loss": 0.6644,
+      "step": 833
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8154425612052736e-05,
+      "loss": 0.6155,
+      "step": 834
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.815207156308851e-05,
+      "loss": 0.5561,
+      "step": 835
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.81497175141243e-05,
+      "loss": 0.6128,
+      "step": 836
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.814736346516008e-05,
+      "loss": 0.6028,
+      "step": 837
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.814500941619586e-05,
+      "loss": 0.6076,
+      "step": 838
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.814265536723164e-05,
+      "loss": 0.5775,
+      "step": 839
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.814030131826742e-05,
+      "loss": 0.6929,
+      "step": 840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.81379472693032e-05,
+      "loss": 0.6453,
+      "step": 841
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.813559322033899e-05,
+      "loss": 0.5984,
+      "step": 842
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8133239171374764e-05,
+      "loss": 0.6715,
+      "step": 843
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.813088512241055e-05,
+      "loss": 0.6081,
+      "step": 844
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.812853107344633e-05,
+      "loss": 0.7005,
+      "step": 845
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.812617702448211e-05,
+      "loss": 0.575,
+      "step": 846
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.812382297551789e-05,
+      "loss": 0.6672,
+      "step": 847
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.812146892655368e-05,
+      "loss": 0.6999,
+      "step": 848
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8119114877589454e-05,
+      "loss": 0.6442,
+      "step": 849
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.811676082862524e-05,
+      "loss": 0.5975,
+      "step": 850
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.811440677966102e-05,
+      "loss": 0.5366,
+      "step": 851
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.81120527306968e-05,
+      "loss": 0.6808,
+      "step": 852
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.810969868173258e-05,
+      "loss": 0.6133,
+      "step": 853
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.810734463276837e-05,
+      "loss": 0.638,
+      "step": 854
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8104990583804144e-05,
+      "loss": 0.6319,
+      "step": 855
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.810263653483993e-05,
+      "loss": 0.6051,
+      "step": 856
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.810028248587571e-05,
+      "loss": 0.6146,
+      "step": 857
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.809792843691149e-05,
+      "loss": 0.6383,
+      "step": 858
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.809557438794727e-05,
+      "loss": 0.5958,
+      "step": 859
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.809322033898305e-05,
+      "loss": 0.5874,
+      "step": 860
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8090866290018834e-05,
+      "loss": 0.5585,
+      "step": 861
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.808851224105462e-05,
+      "loss": 0.6077,
+      "step": 862
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8086158192090395e-05,
+      "loss": 0.5857,
+      "step": 863
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.808380414312618e-05,
+      "loss": 0.6555,
+      "step": 864
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.808145009416196e-05,
+      "loss": 0.5801,
+      "step": 865
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.807909604519774e-05,
+      "loss": 0.6116,
+      "step": 866
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8076741996233524e-05,
+      "loss": 0.6502,
+      "step": 867
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.807438794726931e-05,
+      "loss": 0.6122,
+      "step": 868
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8072033898305085e-05,
+      "loss": 0.6623,
+      "step": 869
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.806967984934087e-05,
+      "loss": 0.5932,
+      "step": 870
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.806732580037665e-05,
+      "loss": 0.5905,
+      "step": 871
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.806497175141243e-05,
+      "loss": 0.5558,
+      "step": 872
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8062617702448214e-05,
+      "loss": 0.5396,
+      "step": 873
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8060263653484e-05,
+      "loss": 0.5407,
+      "step": 874
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8057909604519775e-05,
+      "loss": 0.6221,
+      "step": 875
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.805555555555556e-05,
+      "loss": 0.5977,
+      "step": 876
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8053201506591336e-05,
+      "loss": 0.5898,
+      "step": 877
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.805084745762712e-05,
+      "loss": 0.5895,
+      "step": 878
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8048493408662904e-05,
+      "loss": 0.6666,
+      "step": 879
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.804613935969868e-05,
+      "loss": 0.5937,
+      "step": 880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8043785310734465e-05,
+      "loss": 0.5408,
+      "step": 881
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.804143126177025e-05,
+      "loss": 0.6329,
+      "step": 882
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8039077212806026e-05,
+      "loss": 0.5777,
+      "step": 883
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.803672316384181e-05,
+      "loss": 0.6265,
+      "step": 884
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8034369114877594e-05,
+      "loss": 0.6415,
+      "step": 885
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.803201506591337e-05,
+      "loss": 0.5785,
+      "step": 886
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8029661016949155e-05,
+      "loss": 0.6226,
+      "step": 887
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.802730696798494e-05,
+      "loss": 0.5555,
+      "step": 888
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8024952919020716e-05,
+      "loss": 0.5895,
+      "step": 889
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.80225988700565e-05,
+      "loss": 0.5894,
+      "step": 890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8020244821092284e-05,
+      "loss": 0.6404,
+      "step": 891
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.801789077212806e-05,
+      "loss": 0.6507,
+      "step": 892
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8015536723163845e-05,
+      "loss": 0.6383,
+      "step": 893
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.801318267419962e-05,
+      "loss": 0.5818,
+      "step": 894
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8010828625235406e-05,
+      "loss": 0.6624,
+      "step": 895
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.800847457627119e-05,
+      "loss": 0.605,
+      "step": 896
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.800612052730697e-05,
+      "loss": 0.5405,
+      "step": 897
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.800376647834275e-05,
+      "loss": 0.6288,
+      "step": 898
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8001412429378535e-05,
+      "loss": 0.5494,
+      "step": 899
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.799905838041431e-05,
+      "loss": 0.5406,
+      "step": 900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7996704331450096e-05,
+      "loss": 0.5977,
+      "step": 901
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.799435028248588e-05,
+      "loss": 0.5726,
+      "step": 902
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.799199623352166e-05,
+      "loss": 0.6163,
+      "step": 903
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.798964218455744e-05,
+      "loss": 0.5651,
+      "step": 904
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7987288135593225e-05,
+      "loss": 0.556,
+      "step": 905
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7984934086629e-05,
+      "loss": 0.64,
+      "step": 906
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7982580037664786e-05,
+      "loss": 0.5121,
+      "step": 907
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.798022598870057e-05,
+      "loss": 0.6552,
+      "step": 908
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.797787193973635e-05,
+      "loss": 0.5757,
+      "step": 909
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.797551789077213e-05,
+      "loss": 0.6891,
+      "step": 910
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.797316384180791e-05,
+      "loss": 0.573,
+      "step": 911
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.79708097928437e-05,
+      "loss": 0.6991,
+      "step": 912
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7968455743879476e-05,
+      "loss": 0.5743,
+      "step": 913
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.796610169491525e-05,
+      "loss": 0.7137,
+      "step": 914
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.796374764595104e-05,
+      "loss": 0.5736,
+      "step": 915
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.796139359698682e-05,
+      "loss": 0.5251,
+      "step": 916
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.79590395480226e-05,
+      "loss": 0.617,
+      "step": 917
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.795668549905838e-05,
+      "loss": 0.5703,
+      "step": 918
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7954331450094166e-05,
+      "loss": 0.6639,
+      "step": 919
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.795197740112994e-05,
+      "loss": 0.6415,
+      "step": 920
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.794962335216573e-05,
+      "loss": 0.5932,
+      "step": 921
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.794726930320151e-05,
+      "loss": 0.5338,
+      "step": 922
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.794491525423729e-05,
+      "loss": 0.5875,
+      "step": 923
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.794256120527307e-05,
+      "loss": 0.5926,
+      "step": 924
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7940207156308856e-05,
+      "loss": 0.5815,
+      "step": 925
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.793785310734463e-05,
+      "loss": 0.5715,
+      "step": 926
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.793549905838042e-05,
+      "loss": 0.5369,
+      "step": 927
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7933145009416194e-05,
+      "loss": 0.595,
+      "step": 928
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7930790960451985e-05,
+      "loss": 0.5691,
+      "step": 929
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.792843691148776e-05,
+      "loss": 0.6029,
+      "step": 930
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.792608286252354e-05,
+      "loss": 0.6489,
+      "step": 931
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.792372881355933e-05,
+      "loss": 0.5891,
+      "step": 932
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.792137476459511e-05,
+      "loss": 0.5288,
+      "step": 933
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7919020715630884e-05,
+      "loss": 0.6305,
+      "step": 934
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.791666666666667e-05,
+      "loss": 0.5949,
+      "step": 935
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.791431261770245e-05,
+      "loss": 0.6451,
+      "step": 936
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.791195856873823e-05,
+      "loss": 0.5777,
+      "step": 937
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.790960451977401e-05,
+      "loss": 0.5907,
+      "step": 938
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.79072504708098e-05,
+      "loss": 0.5777,
+      "step": 939
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7904896421845574e-05,
+      "loss": 0.5619,
+      "step": 940
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.790254237288136e-05,
+      "loss": 0.5803,
+      "step": 941
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.790018832391714e-05,
+      "loss": 0.5947,
+      "step": 942
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7897834274952926e-05,
+      "loss": 0.4791,
+      "step": 943
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.78954802259887e-05,
+      "loss": 0.6261,
+      "step": 944
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.789312617702448e-05,
+      "loss": 0.6609,
+      "step": 945
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.789077212806027e-05,
+      "loss": 0.5182,
+      "step": 946
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.788841807909605e-05,
+      "loss": 0.5193,
+      "step": 947
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7886064030131825e-05,
+      "loss": 0.5374,
+      "step": 948
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7883709981167616e-05,
+      "loss": 0.589,
+      "step": 949
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.788135593220339e-05,
+      "loss": 0.7054,
+      "step": 950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.787900188323917e-05,
+      "loss": 0.5411,
+      "step": 951
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7876647834274954e-05,
+      "loss": 0.6457,
+      "step": 952
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.787429378531074e-05,
+      "loss": 0.634,
+      "step": 953
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7871939736346515e-05,
+      "loss": 0.6506,
+      "step": 954
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.78695856873823e-05,
+      "loss": 0.605,
+      "step": 955
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.786723163841808e-05,
+      "loss": 0.6222,
+      "step": 956
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.786487758945386e-05,
+      "loss": 0.4944,
+      "step": 957
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7862523540489644e-05,
+      "loss": 0.6237,
+      "step": 958
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.786016949152543e-05,
+      "loss": 0.5886,
+      "step": 959
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.785781544256121e-05,
+      "loss": 0.6323,
+      "step": 960
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.785546139359699e-05,
+      "loss": 0.6215,
+      "step": 961
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7853107344632766e-05,
+      "loss": 0.6327,
+      "step": 962
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.785075329566856e-05,
+      "loss": 0.6488,
+      "step": 963
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7848399246704334e-05,
+      "loss": 0.615,
+      "step": 964
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.784604519774011e-05,
+      "loss": 0.6327,
+      "step": 965
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.78436911487759e-05,
+      "loss": 0.6419,
+      "step": 966
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.784133709981168e-05,
+      "loss": 0.5815,
+      "step": 967
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7838983050847456e-05,
+      "loss": 0.5937,
+      "step": 968
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.783662900188324e-05,
+      "loss": 0.6221,
+      "step": 969
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7834274952919024e-05,
+      "loss": 0.6163,
+      "step": 970
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.78319209039548e-05,
+      "loss": 0.6384,
+      "step": 971
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7829566854990585e-05,
+      "loss": 0.6095,
+      "step": 972
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.782721280602637e-05,
+      "loss": 0.6176,
+      "step": 973
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7824858757062146e-05,
+      "loss": 0.5537,
+      "step": 974
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.782250470809793e-05,
+      "loss": 0.5937,
+      "step": 975
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7820150659133714e-05,
+      "loss": 0.5471,
+      "step": 976
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.78177966101695e-05,
+      "loss": 0.5661,
+      "step": 977
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7815442561205275e-05,
+      "loss": 0.5721,
+      "step": 978
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.781308851224105e-05,
+      "loss": 0.5541,
+      "step": 979
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.781073446327684e-05,
+      "loss": 0.5174,
+      "step": 980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.780838041431262e-05,
+      "loss": 0.6759,
+      "step": 981
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.78060263653484e-05,
+      "loss": 0.5837,
+      "step": 982
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.780367231638419e-05,
+      "loss": 0.5483,
+      "step": 983
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7801318267419965e-05,
+      "loss": 0.6631,
+      "step": 984
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.779896421845574e-05,
+      "loss": 0.576,
+      "step": 985
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7796610169491526e-05,
+      "loss": 0.6056,
+      "step": 986
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.779425612052731e-05,
+      "loss": 0.5654,
+      "step": 987
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.779190207156309e-05,
+      "loss": 0.5246,
+      "step": 988
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.778954802259887e-05,
+      "loss": 0.6072,
+      "step": 989
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7787193973634655e-05,
+      "loss": 0.6549,
+      "step": 990
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.778483992467044e-05,
+      "loss": 0.5565,
+      "step": 991
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7782485875706216e-05,
+      "loss": 0.5686,
+      "step": 992
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7780131826742e-05,
+      "loss": 0.6389,
+      "step": 993
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7777777777777784e-05,
+      "loss": 0.5548,
+      "step": 994
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.777542372881356e-05,
+      "loss": 0.5707,
+      "step": 995
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.777306967984934e-05,
+      "loss": 0.6854,
+      "step": 996
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.777071563088513e-05,
+      "loss": 0.5947,
+      "step": 997
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7768361581920906e-05,
+      "loss": 0.5621,
+      "step": 998
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7766007532956683e-05,
+      "loss": 0.6049,
+      "step": 999
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7763653483992474e-05,
+      "loss": 0.5892,
+      "step": 1000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.776129943502825e-05,
+      "loss": 0.5727,
+      "step": 1001
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.775894538606403e-05,
+      "loss": 0.7136,
+      "step": 1002
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.775659133709981e-05,
+      "loss": 0.6435,
+      "step": 1003
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7754237288135596e-05,
+      "loss": 0.6543,
+      "step": 1004
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7751883239171373e-05,
+      "loss": 0.6426,
+      "step": 1005
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.774952919020716e-05,
+      "loss": 0.6012,
+      "step": 1006
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.774717514124294e-05,
+      "loss": 0.562,
+      "step": 1007
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7744821092278725e-05,
+      "loss": 0.5602,
+      "step": 1008
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.77424670433145e-05,
+      "loss": 0.5895,
+      "step": 1009
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7740112994350286e-05,
+      "loss": 0.5967,
+      "step": 1010
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.773775894538607e-05,
+      "loss": 0.7087,
+      "step": 1011
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.773540489642185e-05,
+      "loss": 0.5687,
+      "step": 1012
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7733050847457624e-05,
+      "loss": 0.631,
+      "step": 1013
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7730696798493415e-05,
+      "loss": 0.4818,
+      "step": 1014
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.772834274952919e-05,
+      "loss": 0.5714,
+      "step": 1015
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.772598870056497e-05,
+      "loss": 0.6364,
+      "step": 1016
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.772363465160076e-05,
+      "loss": 0.5895,
+      "step": 1017
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.772128060263654e-05,
+      "loss": 0.5855,
+      "step": 1018
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7718926553672314e-05,
+      "loss": 0.5864,
+      "step": 1019
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.77165725047081e-05,
+      "loss": 0.5809,
+      "step": 1020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.771421845574388e-05,
+      "loss": 0.6844,
+      "step": 1021
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7711864406779666e-05,
+      "loss": 0.5592,
+      "step": 1022
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7709510357815443e-05,
+      "loss": 0.6002,
+      "step": 1023
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.770715630885123e-05,
+      "loss": 0.5927,
+      "step": 1024
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.770480225988701e-05,
+      "loss": 0.5838,
+      "step": 1025
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.770244821092279e-05,
+      "loss": 0.6362,
+      "step": 1026
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.770009416195857e-05,
+      "loss": 0.7157,
+      "step": 1027
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7697740112994356e-05,
+      "loss": 0.563,
+      "step": 1028
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7695386064030133e-05,
+      "loss": 0.5688,
+      "step": 1029
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.769303201506591e-05,
+      "loss": 0.5569,
+      "step": 1030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.76906779661017e-05,
+      "loss": 0.5747,
+      "step": 1031
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.768832391713748e-05,
+      "loss": 0.602,
+      "step": 1032
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7685969868173256e-05,
+      "loss": 0.6086,
+      "step": 1033
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7683615819209046e-05,
+      "loss": 0.6181,
+      "step": 1034
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7681261770244823e-05,
+      "loss": 0.5735,
+      "step": 1035
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.76789077212806e-05,
+      "loss": 0.5767,
+      "step": 1036
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7676553672316384e-05,
+      "loss": 0.5419,
+      "step": 1037
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.767419962335217e-05,
+      "loss": 0.7051,
+      "step": 1038
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.767184557438795e-05,
+      "loss": 0.653,
+      "step": 1039
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.766949152542373e-05,
+      "loss": 0.549,
+      "step": 1040
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.766713747645951e-05,
+      "loss": 0.7132,
+      "step": 1041
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.76647834274953e-05,
+      "loss": 0.564,
+      "step": 1042
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7662429378531074e-05,
+      "loss": 0.6448,
+      "step": 1043
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.766007532956686e-05,
+      "loss": 0.6136,
+      "step": 1044
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.765772128060264e-05,
+      "loss": 0.6733,
+      "step": 1045
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.765536723163842e-05,
+      "loss": 0.6189,
+      "step": 1046
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7653013182674197e-05,
+      "loss": 0.6634,
+      "step": 1047
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.765065913370999e-05,
+      "loss": 0.5661,
+      "step": 1048
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7648305084745764e-05,
+      "loss": 0.5251,
+      "step": 1049
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.764595103578154e-05,
+      "loss": 0.608,
+      "step": 1050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.764359698681733e-05,
+      "loss": 0.6729,
+      "step": 1051
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.764124293785311e-05,
+      "loss": 0.6342,
+      "step": 1052
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7638888888888887e-05,
+      "loss": 0.5823,
+      "step": 1053
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.763653483992467e-05,
+      "loss": 0.5972,
+      "step": 1054
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7634180790960454e-05,
+      "loss": 0.6726,
+      "step": 1055
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.763182674199624e-05,
+      "loss": 0.622,
+      "step": 1056
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7629472693032016e-05,
+      "loss": 0.6376,
+      "step": 1057
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.76271186440678e-05,
+      "loss": 0.6049,
+      "step": 1058
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.762476459510358e-05,
+      "loss": 0.6477,
+      "step": 1059
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.762241054613936e-05,
+      "loss": 0.5329,
+      "step": 1060
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7620056497175144e-05,
+      "loss": 0.613,
+      "step": 1061
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.761770244821093e-05,
+      "loss": 0.6095,
+      "step": 1062
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7615348399246706e-05,
+      "loss": 0.5741,
+      "step": 1063
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.761299435028248e-05,
+      "loss": 0.6439,
+      "step": 1064
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.761064030131827e-05,
+      "loss": 0.6045,
+      "step": 1065
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.760828625235405e-05,
+      "loss": 0.5435,
+      "step": 1066
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.760593220338983e-05,
+      "loss": 0.5276,
+      "step": 1067
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.760357815442562e-05,
+      "loss": 0.6162,
+      "step": 1068
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7601224105461396e-05,
+      "loss": 0.6046,
+      "step": 1069
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.759887005649718e-05,
+      "loss": 0.529,
+      "step": 1070
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7596516007532957e-05,
+      "loss": 0.5744,
+      "step": 1071
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.759416195856874e-05,
+      "loss": 0.6524,
+      "step": 1072
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7591807909604524e-05,
+      "loss": 0.5446,
+      "step": 1073
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.75894538606403e-05,
+      "loss": 0.5975,
+      "step": 1074
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7587099811676086e-05,
+      "loss": 0.5597,
+      "step": 1075
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.758474576271187e-05,
+      "loss": 0.6786,
+      "step": 1076
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7582391713747647e-05,
+      "loss": 0.6574,
+      "step": 1077
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.758003766478343e-05,
+      "loss": 0.5852,
+      "step": 1078
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7577683615819214e-05,
+      "loss": 0.512,
+      "step": 1079
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.757532956685499e-05,
+      "loss": 0.6053,
+      "step": 1080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7572975517890775e-05,
+      "loss": 0.5295,
+      "step": 1081
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.757062146892656e-05,
+      "loss": 0.6789,
+      "step": 1082
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7568267419962337e-05,
+      "loss": 0.6406,
+      "step": 1083
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7565913370998114e-05,
+      "loss": 0.5585,
+      "step": 1084
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7563559322033904e-05,
+      "loss": 0.5694,
+      "step": 1085
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.756120527306968e-05,
+      "loss": 0.566,
+      "step": 1086
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7558851224105465e-05,
+      "loss": 0.6614,
+      "step": 1087
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.755649717514124e-05,
+      "loss": 0.537,
+      "step": 1088
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7554143126177027e-05,
+      "loss": 0.6308,
+      "step": 1089
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.755178907721281e-05,
+      "loss": 0.6308,
+      "step": 1090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.754943502824859e-05,
+      "loss": 0.554,
+      "step": 1091
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.754708097928437e-05,
+      "loss": 0.5355,
+      "step": 1092
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7544726930320155e-05,
+      "loss": 0.5546,
+      "step": 1093
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.754237288135593e-05,
+      "loss": 0.5967,
+      "step": 1094
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7540018832391717e-05,
+      "loss": 0.5931,
+      "step": 1095
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.75376647834275e-05,
+      "loss": 0.5328,
+      "step": 1096
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.753531073446328e-05,
+      "loss": 0.7418,
+      "step": 1097
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.753295668549906e-05,
+      "loss": 0.515,
+      "step": 1098
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7530602636534845e-05,
+      "loss": 0.6674,
+      "step": 1099
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.752824858757062e-05,
+      "loss": 0.4829,
+      "step": 1100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7525894538606407e-05,
+      "loss": 0.5905,
+      "step": 1101
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.752354048964219e-05,
+      "loss": 0.5042,
+      "step": 1102
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.752118644067797e-05,
+      "loss": 0.6387,
+      "step": 1103
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.751883239171375e-05,
+      "loss": 0.4897,
+      "step": 1104
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.751647834274953e-05,
+      "loss": 0.5269,
+      "step": 1105
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.751412429378531e-05,
+      "loss": 0.5614,
+      "step": 1106
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7511770244821097e-05,
+      "loss": 0.6766,
+      "step": 1107
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7509416195856874e-05,
+      "loss": 0.6215,
+      "step": 1108
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.750706214689266e-05,
+      "loss": 0.7274,
+      "step": 1109
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.750470809792844e-05,
+      "loss": 0.5437,
+      "step": 1110
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.750235404896422e-05,
+      "loss": 0.6122,
+      "step": 1111
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.75e-05,
+      "loss": 0.6177,
+      "step": 1112
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7497645951035787e-05,
+      "loss": 0.576,
+      "step": 1113
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7495291902071564e-05,
+      "loss": 0.5678,
+      "step": 1114
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.749293785310735e-05,
+      "loss": 0.7428,
+      "step": 1115
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.749058380414313e-05,
+      "loss": 0.5649,
+      "step": 1116
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.748822975517891e-05,
+      "loss": 0.6778,
+      "step": 1117
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.748587570621469e-05,
+      "loss": 0.5645,
+      "step": 1118
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7483521657250477e-05,
+      "loss": 0.5718,
+      "step": 1119
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7481167608286254e-05,
+      "loss": 0.5896,
+      "step": 1120
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.747881355932204e-05,
+      "loss": 0.6049,
+      "step": 1121
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7476459510357815e-05,
+      "loss": 0.6298,
+      "step": 1122
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.74741054613936e-05,
+      "loss": 0.6829,
+      "step": 1123
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.747175141242938e-05,
+      "loss": 0.5639,
+      "step": 1124
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.746939736346516e-05,
+      "loss": 0.5878,
+      "step": 1125
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7467043314500944e-05,
+      "loss": 0.6381,
+      "step": 1126
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.746468926553673e-05,
+      "loss": 0.6003,
+      "step": 1127
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7462335216572505e-05,
+      "loss": 0.6087,
+      "step": 1128
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.745998116760829e-05,
+      "loss": 0.5633,
+      "step": 1129
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.745762711864407e-05,
+      "loss": 0.5612,
+      "step": 1130
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.745527306967985e-05,
+      "loss": 0.5216,
+      "step": 1131
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7452919020715634e-05,
+      "loss": 0.5991,
+      "step": 1132
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.745056497175142e-05,
+      "loss": 0.5151,
+      "step": 1133
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7448210922787195e-05,
+      "loss": 0.5961,
+      "step": 1134
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.744585687382298e-05,
+      "loss": 0.6011,
+      "step": 1135
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.744350282485876e-05,
+      "loss": 0.5484,
+      "step": 1136
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.744114877589454e-05,
+      "loss": 0.5683,
+      "step": 1137
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7438794726930324e-05,
+      "loss": 0.5573,
+      "step": 1138
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.74364406779661e-05,
+      "loss": 0.5757,
+      "step": 1139
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7434086629001885e-05,
+      "loss": 0.676,
+      "step": 1140
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.743173258003767e-05,
+      "loss": 0.5208,
+      "step": 1141
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7429378531073446e-05,
+      "loss": 0.4764,
+      "step": 1142
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.742702448210923e-05,
+      "loss": 0.5856,
+      "step": 1143
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7424670433145014e-05,
+      "loss": 0.584,
+      "step": 1144
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.742231638418079e-05,
+      "loss": 0.5478,
+      "step": 1145
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7419962335216575e-05,
+      "loss": 0.5687,
+      "step": 1146
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.741760828625236e-05,
+      "loss": 0.6162,
+      "step": 1147
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7415254237288136e-05,
+      "loss": 0.638,
+      "step": 1148
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.741290018832392e-05,
+      "loss": 0.5485,
+      "step": 1149
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7410546139359704e-05,
+      "loss": 0.536,
+      "step": 1150
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.740819209039548e-05,
+      "loss": 0.5246,
+      "step": 1151
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7405838041431265e-05,
+      "loss": 0.5695,
+      "step": 1152
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.740348399246705e-05,
+      "loss": 0.5687,
+      "step": 1153
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7401129943502826e-05,
+      "loss": 0.5373,
+      "step": 1154
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.739877589453861e-05,
+      "loss": 0.5978,
+      "step": 1155
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.739642184557439e-05,
+      "loss": 0.5824,
+      "step": 1156
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.739406779661017e-05,
+      "loss": 0.6675,
+      "step": 1157
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7391713747645955e-05,
+      "loss": 0.5365,
+      "step": 1158
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.738935969868173e-05,
+      "loss": 0.5596,
+      "step": 1159
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7387005649717516e-05,
+      "loss": 0.668,
+      "step": 1160
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.73846516007533e-05,
+      "loss": 0.639,
+      "step": 1161
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.738229755178908e-05,
+      "loss": 0.5796,
+      "step": 1162
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.737994350282486e-05,
+      "loss": 0.5954,
+      "step": 1163
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7377589453860645e-05,
+      "loss": 0.5968,
+      "step": 1164
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.737523540489642e-05,
+      "loss": 0.5679,
+      "step": 1165
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7372881355932206e-05,
+      "loss": 0.5627,
+      "step": 1166
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.737052730696799e-05,
+      "loss": 0.5831,
+      "step": 1167
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.736817325800377e-05,
+      "loss": 0.558,
+      "step": 1168
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.736581920903955e-05,
+      "loss": 0.5667,
+      "step": 1169
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7363465160075335e-05,
+      "loss": 0.6021,
+      "step": 1170
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.736111111111111e-05,
+      "loss": 0.5443,
+      "step": 1171
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7358757062146896e-05,
+      "loss": 0.5815,
+      "step": 1172
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.735640301318268e-05,
+      "loss": 0.5241,
+      "step": 1173
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.735404896421846e-05,
+      "loss": 0.6885,
+      "step": 1174
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.735169491525424e-05,
+      "loss": 0.6281,
+      "step": 1175
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.734934086629002e-05,
+      "loss": 0.5531,
+      "step": 1176
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.73469868173258e-05,
+      "loss": 0.6012,
+      "step": 1177
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7344632768361586e-05,
+      "loss": 0.5027,
+      "step": 1178
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.734227871939736e-05,
+      "loss": 0.6676,
+      "step": 1179
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.733992467043315e-05,
+      "loss": 0.5181,
+      "step": 1180
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.733757062146893e-05,
+      "loss": 0.5288,
+      "step": 1181
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.733521657250471e-05,
+      "loss": 0.636,
+      "step": 1182
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.733286252354049e-05,
+      "loss": 0.5702,
+      "step": 1183
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7330508474576276e-05,
+      "loss": 0.5442,
+      "step": 1184
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.732815442561205e-05,
+      "loss": 0.5597,
+      "step": 1185
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.732580037664784e-05,
+      "loss": 0.5684,
+      "step": 1186
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.732344632768362e-05,
+      "loss": 0.6185,
+      "step": 1187
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.73210922787194e-05,
+      "loss": 0.566,
+      "step": 1188
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.731873822975518e-05,
+      "loss": 0.5888,
+      "step": 1189
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7316384180790966e-05,
+      "loss": 0.634,
+      "step": 1190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.731403013182674e-05,
+      "loss": 0.6445,
+      "step": 1191
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.731167608286253e-05,
+      "loss": 0.589,
+      "step": 1192
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7309322033898304e-05,
+      "loss": 0.5821,
+      "step": 1193
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.730696798493409e-05,
+      "loss": 0.6936,
+      "step": 1194
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.730461393596987e-05,
+      "loss": 0.5508,
+      "step": 1195
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.730225988700565e-05,
+      "loss": 0.6465,
+      "step": 1196
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.729990583804143e-05,
+      "loss": 0.5483,
+      "step": 1197
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.729755178907722e-05,
+      "loss": 0.6449,
+      "step": 1198
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7295197740112994e-05,
+      "loss": 0.5852,
+      "step": 1199
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.729284369114878e-05,
+      "loss": 0.6646,
+      "step": 1200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.729048964218456e-05,
+      "loss": 0.5391,
+      "step": 1201
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.728813559322034e-05,
+      "loss": 0.5553,
+      "step": 1202
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.728578154425612e-05,
+      "loss": 0.7029,
+      "step": 1203
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.728342749529191e-05,
+      "loss": 0.5199,
+      "step": 1204
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7281073446327684e-05,
+      "loss": 0.5793,
+      "step": 1205
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.727871939736347e-05,
+      "loss": 0.6098,
+      "step": 1206
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.727636534839925e-05,
+      "loss": 0.6757,
+      "step": 1207
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.727401129943503e-05,
+      "loss": 0.6629,
+      "step": 1208
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.727165725047081e-05,
+      "loss": 0.546,
+      "step": 1209
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.726930320150659e-05,
+      "loss": 0.5626,
+      "step": 1210
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7266949152542374e-05,
+      "loss": 0.6212,
+      "step": 1211
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.726459510357816e-05,
+      "loss": 0.5641,
+      "step": 1212
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7262241054613935e-05,
+      "loss": 0.5771,
+      "step": 1213
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.725988700564972e-05,
+      "loss": 0.6437,
+      "step": 1214
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.72575329566855e-05,
+      "loss": 0.5391,
+      "step": 1215
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.725517890772128e-05,
+      "loss": 0.5344,
+      "step": 1216
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7252824858757064e-05,
+      "loss": 0.5599,
+      "step": 1217
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.725047080979285e-05,
+      "loss": 0.6337,
+      "step": 1218
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7248116760828625e-05,
+      "loss": 0.5976,
+      "step": 1219
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.724576271186441e-05,
+      "loss": 0.6028,
+      "step": 1220
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.724340866290019e-05,
+      "loss": 0.5056,
+      "step": 1221
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.724105461393597e-05,
+      "loss": 0.492,
+      "step": 1222
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7238700564971754e-05,
+      "loss": 0.5863,
+      "step": 1223
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.723634651600754e-05,
+      "loss": 0.5092,
+      "step": 1224
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7233992467043315e-05,
+      "loss": 0.5164,
+      "step": 1225
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.72316384180791e-05,
+      "loss": 0.7087,
+      "step": 1226
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7229284369114876e-05,
+      "loss": 0.6225,
+      "step": 1227
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.722693032015067e-05,
+      "loss": 0.6362,
+      "step": 1228
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7224576271186444e-05,
+      "loss": 0.7482,
+      "step": 1229
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.722222222222222e-05,
+      "loss": 0.6262,
+      "step": 1230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7219868173258005e-05,
+      "loss": 0.5321,
+      "step": 1231
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.721751412429379e-05,
+      "loss": 0.491,
+      "step": 1232
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7215160075329566e-05,
+      "loss": 0.5184,
+      "step": 1233
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.721280602636535e-05,
+      "loss": 0.6358,
+      "step": 1234
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7210451977401134e-05,
+      "loss": 0.6103,
+      "step": 1235
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.720809792843691e-05,
+      "loss": 0.5987,
+      "step": 1236
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7205743879472695e-05,
+      "loss": 0.5571,
+      "step": 1237
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.720338983050848e-05,
+      "loss": 0.5697,
+      "step": 1238
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7201035781544256e-05,
+      "loss": 0.609,
+      "step": 1239
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.719868173258004e-05,
+      "loss": 0.6095,
+      "step": 1240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7196327683615824e-05,
+      "loss": 0.6695,
+      "step": 1241
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.71939736346516e-05,
+      "loss": 0.6838,
+      "step": 1242
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7191619585687385e-05,
+      "loss": 0.5875,
+      "step": 1243
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.718926553672316e-05,
+      "loss": 0.5736,
+      "step": 1244
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.718691148775895e-05,
+      "loss": 0.5529,
+      "step": 1245
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.718455743879473e-05,
+      "loss": 0.663,
+      "step": 1246
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.718220338983051e-05,
+      "loss": 0.5518,
+      "step": 1247
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.71798493408663e-05,
+      "loss": 0.5669,
+      "step": 1248
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7177495291902075e-05,
+      "loss": 0.6686,
+      "step": 1249
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.717514124293785e-05,
+      "loss": 0.6635,
+      "step": 1250
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7172787193973636e-05,
+      "loss": 0.6225,
+      "step": 1251
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.717043314500942e-05,
+      "loss": 0.6238,
+      "step": 1252
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.71680790960452e-05,
+      "loss": 0.7001,
+      "step": 1253
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.716572504708098e-05,
+      "loss": 0.5673,
+      "step": 1254
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7163370998116765e-05,
+      "loss": 0.4968,
+      "step": 1255
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.716101694915254e-05,
+      "loss": 0.698,
+      "step": 1256
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7158662900188326e-05,
+      "loss": 0.5204,
+      "step": 1257
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.715630885122411e-05,
+      "loss": 0.5709,
+      "step": 1258
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7153954802259894e-05,
+      "loss": 0.5903,
+      "step": 1259
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.715160075329567e-05,
+      "loss": 0.605,
+      "step": 1260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.714924670433145e-05,
+      "loss": 0.6389,
+      "step": 1261
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.714689265536724e-05,
+      "loss": 0.6055,
+      "step": 1262
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7144538606403016e-05,
+      "loss": 0.5545,
+      "step": 1263
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.714218455743879e-05,
+      "loss": 0.5946,
+      "step": 1264
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7139830508474584e-05,
+      "loss": 0.5867,
+      "step": 1265
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.713747645951036e-05,
+      "loss": 0.589,
+      "step": 1266
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.713512241054614e-05,
+      "loss": 0.5086,
+      "step": 1267
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.713276836158192e-05,
+      "loss": 0.6328,
+      "step": 1268
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7130414312617706e-05,
+      "loss": 0.5801,
+      "step": 1269
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.712806026365348e-05,
+      "loss": 0.5185,
+      "step": 1270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.712570621468927e-05,
+      "loss": 0.6443,
+      "step": 1271
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.712335216572505e-05,
+      "loss": 0.5759,
+      "step": 1272
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.712099811676083e-05,
+      "loss": 0.5398,
+      "step": 1273
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.711864406779661e-05,
+      "loss": 0.609,
+      "step": 1274
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7116290018832396e-05,
+      "loss": 0.4944,
+      "step": 1275
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.711393596986818e-05,
+      "loss": 0.5815,
+      "step": 1276
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.711158192090396e-05,
+      "loss": 0.541,
+      "step": 1277
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7109227871939734e-05,
+      "loss": 0.6405,
+      "step": 1278
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7106873822975525e-05,
+      "loss": 0.5635,
+      "step": 1279
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.71045197740113e-05,
+      "loss": 0.6525,
+      "step": 1280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.710216572504708e-05,
+      "loss": 0.5163,
+      "step": 1281
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.709981167608287e-05,
+      "loss": 0.5385,
+      "step": 1282
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.709745762711865e-05,
+      "loss": 0.5569,
+      "step": 1283
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7095103578154424e-05,
+      "loss": 0.6071,
+      "step": 1284
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.709274952919021e-05,
+      "loss": 0.5255,
+      "step": 1285
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.709039548022599e-05,
+      "loss": 0.5857,
+      "step": 1286
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.708804143126177e-05,
+      "loss": 0.5701,
+      "step": 1287
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.708568738229755e-05,
+      "loss": 0.6878,
+      "step": 1288
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.708333333333334e-05,
+      "loss": 0.5412,
+      "step": 1289
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7080979284369114e-05,
+      "loss": 0.6054,
+      "step": 1290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.70786252354049e-05,
+      "loss": 0.6227,
+      "step": 1291
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.707627118644068e-05,
+      "loss": 0.583,
+      "step": 1292
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7073917137476466e-05,
+      "loss": 0.5403,
+      "step": 1293
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.707156308851224e-05,
+      "loss": 0.5507,
+      "step": 1294
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.706920903954802e-05,
+      "loss": 0.5842,
+      "step": 1295
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.706685499058381e-05,
+      "loss": 0.6716,
+      "step": 1296
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.706450094161959e-05,
+      "loss": 0.6659,
+      "step": 1297
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7062146892655365e-05,
+      "loss": 0.5082,
+      "step": 1298
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7059792843691156e-05,
+      "loss": 0.7321,
+      "step": 1299
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.705743879472693e-05,
+      "loss": 0.4672,
+      "step": 1300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.705508474576271e-05,
+      "loss": 0.615,
+      "step": 1301
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7052730696798494e-05,
+      "loss": 0.5899,
+      "step": 1302
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.705037664783428e-05,
+      "loss": 0.6271,
+      "step": 1303
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7048022598870055e-05,
+      "loss": 0.6246,
+      "step": 1304
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.704566854990584e-05,
+      "loss": 0.6171,
+      "step": 1305
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.704331450094162e-05,
+      "loss": 0.6236,
+      "step": 1306
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.704096045197741e-05,
+      "loss": 0.5576,
+      "step": 1307
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7038606403013184e-05,
+      "loss": 0.6073,
+      "step": 1308
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.703625235404897e-05,
+      "loss": 0.729,
+      "step": 1309
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.703389830508475e-05,
+      "loss": 0.5811,
+      "step": 1310
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.703154425612053e-05,
+      "loss": 0.5149,
+      "step": 1311
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.7029190207156306e-05,
+      "loss": 0.6186,
+      "step": 1312
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.70268361581921e-05,
+      "loss": 0.5682,
+      "step": 1313
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.7024482109227874e-05,
+      "loss": 0.5489,
+      "step": 1314
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.702212806026365e-05,
+      "loss": 0.5885,
+      "step": 1315
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.701977401129944e-05,
+      "loss": 0.6545,
+      "step": 1316
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.701741996233522e-05,
+      "loss": 0.5514,
+      "step": 1317
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.7015065913370996e-05,
+      "loss": 0.5942,
+      "step": 1318
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.701271186440678e-05,
+      "loss": 0.6081,
+      "step": 1319
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.7010357815442564e-05,
+      "loss": 0.5668,
+      "step": 1320
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.700800376647834e-05,
+      "loss": 0.5069,
+      "step": 1321
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.7005649717514125e-05,
+      "loss": 0.5465,
+      "step": 1322
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.700329566854991e-05,
+      "loss": 0.5564,
+      "step": 1323
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.700094161958569e-05,
+      "loss": 0.5975,
+      "step": 1324
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.699858757062147e-05,
+      "loss": 0.5838,
+      "step": 1325
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6996233521657254e-05,
+      "loss": 0.6247,
+      "step": 1326
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.699387947269304e-05,
+      "loss": 0.5565,
+      "step": 1327
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6991525423728815e-05,
+      "loss": 0.5917,
+      "step": 1328
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.698917137476459e-05,
+      "loss": 0.5666,
+      "step": 1329
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.698681732580038e-05,
+      "loss": 0.4917,
+      "step": 1330
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.698446327683616e-05,
+      "loss": 0.5886,
+      "step": 1331
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.698210922787194e-05,
+      "loss": 0.5358,
+      "step": 1332
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.697975517890773e-05,
+      "loss": 0.6064,
+      "step": 1333
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6977401129943505e-05,
+      "loss": 0.5263,
+      "step": 1334
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.697504708097928e-05,
+      "loss": 0.5591,
+      "step": 1335
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6972693032015066e-05,
+      "loss": 0.6053,
+      "step": 1336
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.697033898305085e-05,
+      "loss": 0.517,
+      "step": 1337
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6967984934086634e-05,
+      "loss": 0.4913,
+      "step": 1338
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.696563088512241e-05,
+      "loss": 0.5474,
+      "step": 1339
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6963276836158195e-05,
+      "loss": 0.665,
+      "step": 1340
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.696092278719398e-05,
+      "loss": 0.6737,
+      "step": 1341
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6958568738229756e-05,
+      "loss": 0.5278,
+      "step": 1342
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.695621468926554e-05,
+      "loss": 0.6599,
+      "step": 1343
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6953860640301324e-05,
+      "loss": 0.5446,
+      "step": 1344
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.69515065913371e-05,
+      "loss": 0.6459,
+      "step": 1345
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.694915254237288e-05,
+      "loss": 0.527,
+      "step": 1346
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.694679849340867e-05,
+      "loss": 0.6077,
+      "step": 1347
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6944444444444446e-05,
+      "loss": 0.4746,
+      "step": 1348
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6942090395480224e-05,
+      "loss": 0.5136,
+      "step": 1349
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6939736346516014e-05,
+      "loss": 0.6612,
+      "step": 1350
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.693738229755179e-05,
+      "loss": 0.6032,
+      "step": 1351
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.693502824858757e-05,
+      "loss": 0.4972,
+      "step": 1352
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.693267419962335e-05,
+      "loss": 0.5301,
+      "step": 1353
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6930320150659136e-05,
+      "loss": 0.5326,
+      "step": 1354
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.692796610169492e-05,
+      "loss": 0.5338,
+      "step": 1355
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.69256120527307e-05,
+      "loss": 0.6023,
+      "step": 1356
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.692325800376648e-05,
+      "loss": 0.6037,
+      "step": 1357
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6920903954802265e-05,
+      "loss": 0.5917,
+      "step": 1358
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.691854990583804e-05,
+      "loss": 0.623,
+      "step": 1359
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6916195856873826e-05,
+      "loss": 0.5088,
+      "step": 1360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.691384180790961e-05,
+      "loss": 0.5978,
+      "step": 1361
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.691148775894539e-05,
+      "loss": 0.5528,
+      "step": 1362
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6909133709981165e-05,
+      "loss": 0.595,
+      "step": 1363
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6906779661016955e-05,
+      "loss": 0.508,
+      "step": 1364
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.690442561205273e-05,
+      "loss": 0.6445,
+      "step": 1365
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.690207156308851e-05,
+      "loss": 0.5,
+      "step": 1366
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.68997175141243e-05,
+      "loss": 0.6556,
+      "step": 1367
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.689736346516008e-05,
+      "loss": 0.614,
+      "step": 1368
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6895009416195855e-05,
+      "loss": 0.6806,
+      "step": 1369
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.689265536723164e-05,
+      "loss": 0.5394,
+      "step": 1370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.689030131826742e-05,
+      "loss": 0.5802,
+      "step": 1371
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6887947269303206e-05,
+      "loss": 0.557,
+      "step": 1372
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6885593220338983e-05,
+      "loss": 0.5691,
+      "step": 1373
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.688323917137477e-05,
+      "loss": 0.4973,
+      "step": 1374
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.688088512241055e-05,
+      "loss": 0.5939,
+      "step": 1375
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.687853107344633e-05,
+      "loss": 0.6216,
+      "step": 1376
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.687617702448211e-05,
+      "loss": 0.5828,
+      "step": 1377
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6873822975517896e-05,
+      "loss": 0.5673,
+      "step": 1378
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6871468926553673e-05,
+      "loss": 0.5171,
+      "step": 1379
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.686911487758945e-05,
+      "loss": 0.5813,
+      "step": 1380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.686676082862524e-05,
+      "loss": 0.5932,
+      "step": 1381
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.686440677966102e-05,
+      "loss": 0.544,
+      "step": 1382
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6862052730696796e-05,
+      "loss": 0.4942,
+      "step": 1383
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6859698681732586e-05,
+      "loss": 0.5674,
+      "step": 1384
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6857344632768363e-05,
+      "loss": 0.5899,
+      "step": 1385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.685499058380415e-05,
+      "loss": 0.6145,
+      "step": 1386
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6852636534839925e-05,
+      "loss": 0.6462,
+      "step": 1387
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.685028248587571e-05,
+      "loss": 0.6346,
+      "step": 1388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.684792843691149e-05,
+      "loss": 0.5969,
+      "step": 1389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.684557438794727e-05,
+      "loss": 0.5284,
+      "step": 1390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6843220338983053e-05,
+      "loss": 0.5359,
+      "step": 1391
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.684086629001884e-05,
+      "loss": 0.5487,
+      "step": 1392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6838512241054615e-05,
+      "loss": 0.6146,
+      "step": 1393
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.68361581920904e-05,
+      "loss": 0.5824,
+      "step": 1394
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.683380414312618e-05,
+      "loss": 0.5488,
+      "step": 1395
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.683145009416196e-05,
+      "loss": 0.5291,
+      "step": 1396
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6829096045197743e-05,
+      "loss": 0.5399,
+      "step": 1397
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.682674199623353e-05,
+      "loss": 0.6184,
+      "step": 1398
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6824387947269305e-05,
+      "loss": 0.5806,
+      "step": 1399
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.682203389830508e-05,
+      "loss": 0.5305,
+      "step": 1400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.681967984934087e-05,
+      "loss": 0.4339,
+      "step": 1401
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.681732580037665e-05,
+      "loss": 0.6025,
+      "step": 1402
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6814971751412433e-05,
+      "loss": 0.5851,
+      "step": 1403
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.681261770244821e-05,
+      "loss": 0.5213,
+      "step": 1404
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6810263653483995e-05,
+      "loss": 0.6232,
+      "step": 1405
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.680790960451978e-05,
+      "loss": 0.4642,
+      "step": 1406
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6805555555555556e-05,
+      "loss": 0.5419,
+      "step": 1407
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.680320150659134e-05,
+      "loss": 0.5069,
+      "step": 1408
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6800847457627123e-05,
+      "loss": 0.5736,
+      "step": 1409
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.67984934086629e-05,
+      "loss": 0.5456,
+      "step": 1410
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6796139359698685e-05,
+      "loss": 0.5564,
+      "step": 1411
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.679378531073447e-05,
+      "loss": 0.6641,
+      "step": 1412
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6791431261770246e-05,
+      "loss": 0.5577,
+      "step": 1413
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.678907721280603e-05,
+      "loss": 0.5436,
+      "step": 1414
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6786723163841813e-05,
+      "loss": 0.5574,
+      "step": 1415
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.678436911487759e-05,
+      "loss": 0.5577,
+      "step": 1416
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.678201506591337e-05,
+      "loss": 0.7128,
+      "step": 1417
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.677966101694916e-05,
+      "loss": 0.5329,
+      "step": 1418
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6777306967984936e-05,
+      "loss": 0.5772,
+      "step": 1419
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.677495291902072e-05,
+      "loss": 0.5135,
+      "step": 1420
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.67725988700565e-05,
+      "loss": 0.5366,
+      "step": 1421
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.677024482109228e-05,
+      "loss": 0.5643,
+      "step": 1422
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6767890772128065e-05,
+      "loss": 0.5443,
+      "step": 1423
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.676553672316384e-05,
+      "loss": 0.6548,
+      "step": 1424
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6763182674199626e-05,
+      "loss": 0.6396,
+      "step": 1425
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.676082862523541e-05,
+      "loss": 0.5966,
+      "step": 1426
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.675847457627119e-05,
+      "loss": 0.5314,
+      "step": 1427
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.675612052730697e-05,
+      "loss": 0.5912,
+      "step": 1428
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6753766478342755e-05,
+      "loss": 0.5156,
+      "step": 1429
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.675141242937853e-05,
+      "loss": 0.652,
+      "step": 1430
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6749058380414316e-05,
+      "loss": 0.5895,
+      "step": 1431
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.67467043314501e-05,
+      "loss": 0.4965,
+      "step": 1432
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.674435028248588e-05,
+      "loss": 0.5837,
+      "step": 1433
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.674199623352166e-05,
+      "loss": 0.538,
+      "step": 1434
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6739642184557444e-05,
+      "loss": 0.5588,
+      "step": 1435
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.673728813559322e-05,
+      "loss": 0.5677,
+      "step": 1436
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6734934086629006e-05,
+      "loss": 0.5259,
+      "step": 1437
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.673258003766478e-05,
+      "loss": 0.5868,
+      "step": 1438
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.673022598870057e-05,
+      "loss": 0.5405,
+      "step": 1439
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.672787193973635e-05,
+      "loss": 0.4885,
+      "step": 1440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.672551789077213e-05,
+      "loss": 0.582,
+      "step": 1441
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.672316384180791e-05,
+      "loss": 0.5372,
+      "step": 1442
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6720809792843696e-05,
+      "loss": 0.4826,
+      "step": 1443
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.671845574387947e-05,
+      "loss": 0.5186,
+      "step": 1444
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.671610169491526e-05,
+      "loss": 0.5739,
+      "step": 1445
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.671374764595104e-05,
+      "loss": 0.6121,
+      "step": 1446
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.671139359698682e-05,
+      "loss": 0.6453,
+      "step": 1447
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.67090395480226e-05,
+      "loss": 0.5759,
+      "step": 1448
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6706685499058386e-05,
+      "loss": 0.6055,
+      "step": 1449
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.670433145009416e-05,
+      "loss": 0.6069,
+      "step": 1450
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.670197740112995e-05,
+      "loss": 0.588,
+      "step": 1451
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.669962335216573e-05,
+      "loss": 0.5474,
+      "step": 1452
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.669726930320151e-05,
+      "loss": 0.4365,
+      "step": 1453
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.669491525423729e-05,
+      "loss": 0.5833,
+      "step": 1454
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.669256120527307e-05,
+      "loss": 0.5309,
+      "step": 1455
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.669020715630885e-05,
+      "loss": 0.5438,
+      "step": 1456
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.668785310734464e-05,
+      "loss": 0.5377,
+      "step": 1457
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6685499058380414e-05,
+      "loss": 0.5706,
+      "step": 1458
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.66831450094162e-05,
+      "loss": 0.5105,
+      "step": 1459
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.668079096045198e-05,
+      "loss": 0.5592,
+      "step": 1460
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.667843691148776e-05,
+      "loss": 0.5307,
+      "step": 1461
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.667608286252354e-05,
+      "loss": 0.5415,
+      "step": 1462
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.667372881355933e-05,
+      "loss": 0.6803,
+      "step": 1463
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6671374764595104e-05,
+      "loss": 0.6094,
+      "step": 1464
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.666902071563089e-05,
+      "loss": 0.5512,
+      "step": 1465
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.666666666666667e-05,
+      "loss": 0.5559,
+      "step": 1466
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.666431261770245e-05,
+      "loss": 0.6089,
+      "step": 1467
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.666195856873823e-05,
+      "loss": 0.5936,
+      "step": 1468
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6659604519774017e-05,
+      "loss": 0.5317,
+      "step": 1469
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6657250470809794e-05,
+      "loss": 0.5147,
+      "step": 1470
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.665489642184558e-05,
+      "loss": 0.5708,
+      "step": 1471
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6652542372881355e-05,
+      "loss": 0.5279,
+      "step": 1472
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.665018832391714e-05,
+      "loss": 0.5711,
+      "step": 1473
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.664783427495292e-05,
+      "loss": 0.6292,
+      "step": 1474
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.66454802259887e-05,
+      "loss": 0.4902,
+      "step": 1475
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6643126177024484e-05,
+      "loss": 0.5894,
+      "step": 1476
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.664077212806027e-05,
+      "loss": 0.5197,
+      "step": 1477
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6638418079096045e-05,
+      "loss": 0.5927,
+      "step": 1478
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.663606403013183e-05,
+      "loss": 0.4987,
+      "step": 1479
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.663370998116761e-05,
+      "loss": 0.518,
+      "step": 1480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.663135593220339e-05,
+      "loss": 0.5807,
+      "step": 1481
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6629001883239174e-05,
+      "loss": 0.6554,
+      "step": 1482
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.662664783427496e-05,
+      "loss": 0.5715,
+      "step": 1483
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6624293785310735e-05,
+      "loss": 0.5947,
+      "step": 1484
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.662193973634652e-05,
+      "loss": 0.6982,
+      "step": 1485
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.66195856873823e-05,
+      "loss": 0.6181,
+      "step": 1486
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.661723163841808e-05,
+      "loss": 0.5241,
+      "step": 1487
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6614877589453864e-05,
+      "loss": 0.6341,
+      "step": 1488
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.661252354048965e-05,
+      "loss": 0.558,
+      "step": 1489
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6610169491525425e-05,
+      "loss": 0.5709,
+      "step": 1490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.660781544256121e-05,
+      "loss": 0.5886,
+      "step": 1491
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6605461393596986e-05,
+      "loss": 0.538,
+      "step": 1492
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.660310734463277e-05,
+      "loss": 0.6379,
+      "step": 1493
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6600753295668554e-05,
+      "loss": 0.5893,
+      "step": 1494
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.659839924670433e-05,
+      "loss": 0.5824,
+      "step": 1495
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6596045197740115e-05,
+      "loss": 0.7418,
+      "step": 1496
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.65936911487759e-05,
+      "loss": 0.5193,
+      "step": 1497
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6591337099811676e-05,
+      "loss": 0.6861,
+      "step": 1498
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.658898305084746e-05,
+      "loss": 0.4468,
+      "step": 1499
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6586629001883244e-05,
+      "loss": 0.5916,
+      "step": 1500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.658427495291902e-05,
+      "loss": 0.5572,
+      "step": 1501
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6581920903954805e-05,
+      "loss": 0.6361,
+      "step": 1502
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.657956685499059e-05,
+      "loss": 0.5733,
+      "step": 1503
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6577212806026366e-05,
+      "loss": 0.6152,
+      "step": 1504
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.657485875706215e-05,
+      "loss": 0.6057,
+      "step": 1505
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6572504708097934e-05,
+      "loss": 0.592,
+      "step": 1506
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.657015065913371e-05,
+      "loss": 0.5369,
+      "step": 1507
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6567796610169495e-05,
+      "loss": 0.5064,
+      "step": 1508
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.656544256120527e-05,
+      "loss": 0.6483,
+      "step": 1509
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6563088512241056e-05,
+      "loss": 0.6141,
+      "step": 1510
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.656073446327684e-05,
+      "loss": 0.5604,
+      "step": 1511
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.655838041431262e-05,
+      "loss": 0.6031,
+      "step": 1512
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.65560263653484e-05,
+      "loss": 0.5185,
+      "step": 1513
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6553672316384185e-05,
+      "loss": 0.4911,
+      "step": 1514
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.655131826741996e-05,
+      "loss": 0.5682,
+      "step": 1515
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6548964218455746e-05,
+      "loss": 0.5867,
+      "step": 1516
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.654661016949153e-05,
+      "loss": 0.5916,
+      "step": 1517
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.654425612052731e-05,
+      "loss": 0.5794,
+      "step": 1518
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.654190207156309e-05,
+      "loss": 0.4399,
+      "step": 1519
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6539548022598875e-05,
+      "loss": 0.5883,
+      "step": 1520
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.653719397363465e-05,
+      "loss": 0.5694,
+      "step": 1521
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6534839924670436e-05,
+      "loss": 0.553,
+      "step": 1522
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.653248587570622e-05,
+      "loss": 0.5016,
+      "step": 1523
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6530131826742e-05,
+      "loss": 0.5935,
+      "step": 1524
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.652777777777778e-05,
+      "loss": 0.6043,
+      "step": 1525
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.652542372881356e-05,
+      "loss": 0.4603,
+      "step": 1526
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.652306967984934e-05,
+      "loss": 0.5365,
+      "step": 1527
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6520715630885126e-05,
+      "loss": 0.5654,
+      "step": 1528
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.65183615819209e-05,
+      "loss": 0.5401,
+      "step": 1529
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.651600753295669e-05,
+      "loss": 0.5475,
+      "step": 1530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.651365348399247e-05,
+      "loss": 0.4467,
+      "step": 1531
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.651129943502825e-05,
+      "loss": 0.6001,
+      "step": 1532
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.650894538606403e-05,
+      "loss": 0.5331,
+      "step": 1533
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6506591337099816e-05,
+      "loss": 0.4579,
+      "step": 1534
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.650423728813559e-05,
+      "loss": 0.5355,
+      "step": 1535
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.650188323917138e-05,
+      "loss": 0.5141,
+      "step": 1536
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.649952919020716e-05,
+      "loss": 0.5716,
+      "step": 1537
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.649717514124294e-05,
+      "loss": 0.6204,
+      "step": 1538
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.649482109227872e-05,
+      "loss": 0.5519,
+      "step": 1539
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6492467043314506e-05,
+      "loss": 0.5591,
+      "step": 1540
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.649011299435028e-05,
+      "loss": 0.6689,
+      "step": 1541
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.648775894538607e-05,
+      "loss": 0.4384,
+      "step": 1542
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6485404896421844e-05,
+      "loss": 0.5852,
+      "step": 1543
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6483050847457635e-05,
+      "loss": 0.5256,
+      "step": 1544
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.648069679849341e-05,
+      "loss": 0.5894,
+      "step": 1545
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.647834274952919e-05,
+      "loss": 0.5893,
+      "step": 1546
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.647598870056497e-05,
+      "loss": 0.5797,
+      "step": 1547
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.647363465160076e-05,
+      "loss": 0.6019,
+      "step": 1548
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6471280602636534e-05,
+      "loss": 0.6338,
+      "step": 1549
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.646892655367232e-05,
+      "loss": 0.6319,
+      "step": 1550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.64665725047081e-05,
+      "loss": 0.5079,
+      "step": 1551
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.646421845574388e-05,
+      "loss": 0.5741,
+      "step": 1552
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.646186440677966e-05,
+      "loss": 0.5592,
+      "step": 1553
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.645951035781545e-05,
+      "loss": 0.5933,
+      "step": 1554
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6457156308851224e-05,
+      "loss": 0.5668,
+      "step": 1555
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.645480225988701e-05,
+      "loss": 0.6471,
+      "step": 1556
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.645244821092279e-05,
+      "loss": 0.6411,
+      "step": 1557
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.645009416195857e-05,
+      "loss": 0.647,
+      "step": 1558
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.644774011299435e-05,
+      "loss": 0.578,
+      "step": 1559
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.644538606403013e-05,
+      "loss": 0.507,
+      "step": 1560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.644303201506592e-05,
+      "loss": 0.5166,
+      "step": 1561
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.64406779661017e-05,
+      "loss": 0.6174,
+      "step": 1562
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6438323917137475e-05,
+      "loss": 0.5634,
+      "step": 1563
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6435969868173266e-05,
+      "loss": 0.5291,
+      "step": 1564
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.643361581920904e-05,
+      "loss": 0.5993,
+      "step": 1565
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.643126177024482e-05,
+      "loss": 0.562,
+      "step": 1566
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6428907721280604e-05,
+      "loss": 0.5551,
+      "step": 1567
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.642655367231639e-05,
+      "loss": 0.517,
+      "step": 1568
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6424199623352165e-05,
+      "loss": 0.5867,
+      "step": 1569
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.642184557438795e-05,
+      "loss": 0.4842,
+      "step": 1570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.641949152542373e-05,
+      "loss": 0.487,
+      "step": 1571
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.641713747645951e-05,
+      "loss": 0.4189,
+      "step": 1572
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6414783427495294e-05,
+      "loss": 0.5627,
+      "step": 1573
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.641242937853108e-05,
+      "loss": 0.5018,
+      "step": 1574
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6410075329566855e-05,
+      "loss": 0.4888,
+      "step": 1575
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.640772128060264e-05,
+      "loss": 0.6392,
+      "step": 1576
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6405367231638416e-05,
+      "loss": 0.6451,
+      "step": 1577
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.640301318267421e-05,
+      "loss": 0.5864,
+      "step": 1578
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6400659133709984e-05,
+      "loss": 0.446,
+      "step": 1579
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.639830508474576e-05,
+      "loss": 0.5327,
+      "step": 1580
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.639595103578155e-05,
+      "loss": 0.5327,
+      "step": 1581
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.639359698681733e-05,
+      "loss": 0.6918,
+      "step": 1582
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6391242937853106e-05,
+      "loss": 0.5549,
+      "step": 1583
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.638888888888889e-05,
+      "loss": 0.5079,
+      "step": 1584
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6386534839924674e-05,
+      "loss": 0.5911,
+      "step": 1585
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.638418079096045e-05,
+      "loss": 0.5842,
+      "step": 1586
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6381826741996235e-05,
+      "loss": 0.5399,
+      "step": 1587
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.637947269303202e-05,
+      "loss": 0.6261,
+      "step": 1588
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6377118644067796e-05,
+      "loss": 0.4261,
+      "step": 1589
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.637476459510358e-05,
+      "loss": 0.5201,
+      "step": 1590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6372410546139364e-05,
+      "loss": 0.3936,
+      "step": 1591
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.637005649717515e-05,
+      "loss": 0.6301,
+      "step": 1592
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6367702448210925e-05,
+      "loss": 0.6117,
+      "step": 1593
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.63653483992467e-05,
+      "loss": 0.5137,
+      "step": 1594
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.636299435028249e-05,
+      "loss": 0.535,
+      "step": 1595
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.636064030131827e-05,
+      "loss": 0.44,
+      "step": 1596
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.635828625235405e-05,
+      "loss": 0.5099,
+      "step": 1597
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.635593220338984e-05,
+      "loss": 0.4961,
+      "step": 1598
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6353578154425615e-05,
+      "loss": 0.5484,
+      "step": 1599
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.635122410546139e-05,
+      "loss": 0.5525,
+      "step": 1600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6348870056497176e-05,
+      "loss": 0.6866,
+      "step": 1601
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.634651600753296e-05,
+      "loss": 0.5344,
+      "step": 1602
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.634416195856874e-05,
+      "loss": 0.5865,
+      "step": 1603
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.634180790960452e-05,
+      "loss": 0.5989,
+      "step": 1604
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6339453860640305e-05,
+      "loss": 0.5942,
+      "step": 1605
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.633709981167608e-05,
+      "loss": 0.5786,
+      "step": 1606
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6334745762711866e-05,
+      "loss": 0.4678,
+      "step": 1607
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.633239171374765e-05,
+      "loss": 0.5259,
+      "step": 1608
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6330037664783434e-05,
+      "loss": 0.46,
+      "step": 1609
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.632768361581921e-05,
+      "loss": 0.5637,
+      "step": 1610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.632532956685499e-05,
+      "loss": 0.6315,
+      "step": 1611
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.632297551789078e-05,
+      "loss": 0.6605,
+      "step": 1612
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6320621468926556e-05,
+      "loss": 0.5671,
+      "step": 1613
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.631826741996233e-05,
+      "loss": 0.6173,
+      "step": 1614
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6315913370998124e-05,
+      "loss": 0.6155,
+      "step": 1615
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.63135593220339e-05,
+      "loss": 0.5156,
+      "step": 1616
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.631120527306968e-05,
+      "loss": 0.3922,
+      "step": 1617
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.630885122410546e-05,
+      "loss": 0.4846,
+      "step": 1618
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6306497175141246e-05,
+      "loss": 0.6476,
+      "step": 1619
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.630414312617702e-05,
+      "loss": 0.5539,
+      "step": 1620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.630178907721281e-05,
+      "loss": 0.5212,
+      "step": 1621
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.629943502824859e-05,
+      "loss": 0.5484,
+      "step": 1622
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6297080979284375e-05,
+      "loss": 0.5172,
+      "step": 1623
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.629472693032015e-05,
+      "loss": 0.6693,
+      "step": 1624
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6292372881355936e-05,
+      "loss": 0.5385,
+      "step": 1625
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.629001883239172e-05,
+      "loss": 0.6476,
+      "step": 1626
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.62876647834275e-05,
+      "loss": 0.6714,
+      "step": 1627
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6285310734463274e-05,
+      "loss": 0.4936,
+      "step": 1628
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6282956685499065e-05,
+      "loss": 0.5419,
+      "step": 1629
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.628060263653484e-05,
+      "loss": 0.4984,
+      "step": 1630
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.627824858757062e-05,
+      "loss": 0.5665,
+      "step": 1631
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.627589453860641e-05,
+      "loss": 0.4681,
+      "step": 1632
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.627354048964219e-05,
+      "loss": 0.6222,
+      "step": 1633
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6271186440677964e-05,
+      "loss": 0.6264,
+      "step": 1634
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.626883239171375e-05,
+      "loss": 0.6449,
+      "step": 1635
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.626647834274953e-05,
+      "loss": 0.5385,
+      "step": 1636
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.626412429378531e-05,
+      "loss": 0.6181,
+      "step": 1637
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.626177024482109e-05,
+      "loss": 0.7311,
+      "step": 1638
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.625941619585688e-05,
+      "loss": 0.5376,
+      "step": 1639
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.625706214689266e-05,
+      "loss": 0.5836,
+      "step": 1640
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.625470809792844e-05,
+      "loss": 0.5525,
+      "step": 1641
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.625235404896422e-05,
+      "loss": 0.5046,
+      "step": 1642
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6250000000000006e-05,
+      "loss": 0.5452,
+      "step": 1643
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.624764595103578e-05,
+      "loss": 0.5311,
+      "step": 1644
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.624529190207156e-05,
+      "loss": 0.5508,
+      "step": 1645
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.624293785310735e-05,
+      "loss": 0.5678,
+      "step": 1646
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.624058380414313e-05,
+      "loss": 0.4838,
+      "step": 1647
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6238229755178905e-05,
+      "loss": 0.6451,
+      "step": 1648
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6235875706214696e-05,
+      "loss": 0.5771,
+      "step": 1649
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.623352165725047e-05,
+      "loss": 0.5062,
+      "step": 1650
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.623116760828625e-05,
+      "loss": 0.5191,
+      "step": 1651
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6228813559322034e-05,
+      "loss": 0.4248,
+      "step": 1652
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.622645951035782e-05,
+      "loss": 0.5277,
+      "step": 1653
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6224105461393595e-05,
+      "loss": 0.492,
+      "step": 1654
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.622175141242938e-05,
+      "loss": 0.5218,
+      "step": 1655
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.621939736346516e-05,
+      "loss": 0.4534,
+      "step": 1656
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.621704331450095e-05,
+      "loss": 0.5334,
+      "step": 1657
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6214689265536724e-05,
+      "loss": 0.4917,
+      "step": 1658
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.621233521657251e-05,
+      "loss": 0.4903,
+      "step": 1659
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.620998116760829e-05,
+      "loss": 0.5469,
+      "step": 1660
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.620762711864407e-05,
+      "loss": 0.5628,
+      "step": 1661
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6205273069679846e-05,
+      "loss": 0.6198,
+      "step": 1662
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.620291902071564e-05,
+      "loss": 0.6251,
+      "step": 1663
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6200564971751414e-05,
+      "loss": 0.4735,
+      "step": 1664
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.619821092278719e-05,
+      "loss": 0.525,
+      "step": 1665
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.619585687382298e-05,
+      "loss": 0.5486,
+      "step": 1666
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.619350282485876e-05,
+      "loss": 0.6699,
+      "step": 1667
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6191148775894536e-05,
+      "loss": 0.4605,
+      "step": 1668
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.618879472693032e-05,
+      "loss": 0.3896,
+      "step": 1669
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6186440677966104e-05,
+      "loss": 0.4928,
+      "step": 1670
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.618408662900189e-05,
+      "loss": 0.5022,
+      "step": 1671
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6181732580037665e-05,
+      "loss": 0.5689,
+      "step": 1672
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.617937853107345e-05,
+      "loss": 0.5116,
+      "step": 1673
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.617702448210923e-05,
+      "loss": 0.5958,
+      "step": 1674
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.617467043314501e-05,
+      "loss": 0.5797,
+      "step": 1675
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6172316384180794e-05,
+      "loss": 0.4507,
+      "step": 1676
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.616996233521658e-05,
+      "loss": 0.5781,
+      "step": 1677
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6167608286252355e-05,
+      "loss": 0.6061,
+      "step": 1678
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.616525423728813e-05,
+      "loss": 0.5463,
+      "step": 1679
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.616290018832392e-05,
+      "loss": 0.4648,
+      "step": 1680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.61605461393597e-05,
+      "loss": 0.5288,
+      "step": 1681
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.615819209039548e-05,
+      "loss": 0.5572,
+      "step": 1682
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.615583804143127e-05,
+      "loss": 0.5575,
+      "step": 1683
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6153483992467045e-05,
+      "loss": 0.6124,
+      "step": 1684
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.615112994350282e-05,
+      "loss": 0.4943,
+      "step": 1685
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6148775894538606e-05,
+      "loss": 0.5201,
+      "step": 1686
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.614642184557439e-05,
+      "loss": 0.5574,
+      "step": 1687
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6144067796610174e-05,
+      "loss": 0.4204,
+      "step": 1688
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.614171374764595e-05,
+      "loss": 0.5169,
+      "step": 1689
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6139359698681735e-05,
+      "loss": 0.5216,
+      "step": 1690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.613700564971752e-05,
+      "loss": 0.4594,
+      "step": 1691
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6134651600753296e-05,
+      "loss": 0.6046,
+      "step": 1692
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.613229755178908e-05,
+      "loss": 0.5131,
+      "step": 1693
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6129943502824864e-05,
+      "loss": 0.6139,
+      "step": 1694
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.612758945386064e-05,
+      "loss": 0.4761,
+      "step": 1695
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.612523540489642e-05,
+      "loss": 0.5202,
+      "step": 1696
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.612288135593221e-05,
+      "loss": 0.5413,
+      "step": 1697
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6120527306967986e-05,
+      "loss": 0.5485,
+      "step": 1698
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6118173258003764e-05,
+      "loss": 0.601,
+      "step": 1699
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6115819209039554e-05,
+      "loss": 0.5966,
+      "step": 1700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.611346516007533e-05,
+      "loss": 0.4134,
+      "step": 1701
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6111111111111115e-05,
+      "loss": 0.4647,
+      "step": 1702
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.610875706214689e-05,
+      "loss": 0.587,
+      "step": 1703
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6106403013182676e-05,
+      "loss": 0.5461,
+      "step": 1704
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.610404896421846e-05,
+      "loss": 0.6569,
+      "step": 1705
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.610169491525424e-05,
+      "loss": 0.5029,
+      "step": 1706
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.609934086629002e-05,
+      "loss": 0.59,
+      "step": 1707
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6096986817325805e-05,
+      "loss": 0.5142,
+      "step": 1708
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.609463276836158e-05,
+      "loss": 0.6051,
+      "step": 1709
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6092278719397366e-05,
+      "loss": 0.5426,
+      "step": 1710
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.608992467043315e-05,
+      "loss": 0.5218,
+      "step": 1711
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.608757062146893e-05,
+      "loss": 0.5492,
+      "step": 1712
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.608521657250471e-05,
+      "loss": 0.5389,
+      "step": 1713
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6082862523540495e-05,
+      "loss": 0.5264,
+      "step": 1714
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.608050847457627e-05,
+      "loss": 0.5434,
+      "step": 1715
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.607815442561205e-05,
+      "loss": 0.5645,
+      "step": 1716
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.607580037664784e-05,
+      "loss": 0.5727,
+      "step": 1717
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.607344632768362e-05,
+      "loss": 0.515,
+      "step": 1718
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.60710922787194e-05,
+      "loss": 0.5492,
+      "step": 1719
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.606873822975518e-05,
+      "loss": 0.4492,
+      "step": 1720
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.606638418079096e-05,
+      "loss": 0.5724,
+      "step": 1721
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6064030131826746e-05,
+      "loss": 0.5112,
+      "step": 1722
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6061676082862524e-05,
+      "loss": 0.5428,
+      "step": 1723
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.605932203389831e-05,
+      "loss": 0.5246,
+      "step": 1724
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.605696798493409e-05,
+      "loss": 0.5202,
+      "step": 1725
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.605461393596987e-05,
+      "loss": 0.4891,
+      "step": 1726
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.605225988700565e-05,
+      "loss": 0.6684,
+      "step": 1727
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6049905838041436e-05,
+      "loss": 0.5898,
+      "step": 1728
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6047551789077214e-05,
+      "loss": 0.5655,
+      "step": 1729
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6045197740113e-05,
+      "loss": 0.5678,
+      "step": 1730
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.604284369114878e-05,
+      "loss": 0.5412,
+      "step": 1731
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.604048964218456e-05,
+      "loss": 0.503,
+      "step": 1732
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6038135593220336e-05,
+      "loss": 0.5223,
+      "step": 1733
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6035781544256126e-05,
+      "loss": 0.4668,
+      "step": 1734
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6033427495291904e-05,
+      "loss": 0.5438,
+      "step": 1735
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.603107344632769e-05,
+      "loss": 0.5109,
+      "step": 1736
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.6028719397363465e-05,
+      "loss": 0.4781,
+      "step": 1737
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.602636534839925e-05,
+      "loss": 0.5329,
+      "step": 1738
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.602401129943503e-05,
+      "loss": 0.5132,
+      "step": 1739
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.602165725047081e-05,
+      "loss": 0.608,
+      "step": 1740
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.6019303201506594e-05,
+      "loss": 0.5412,
+      "step": 1741
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.601694915254238e-05,
+      "loss": 0.6759,
+      "step": 1742
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.6014595103578155e-05,
+      "loss": 0.5446,
+      "step": 1743
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.601224105461394e-05,
+      "loss": 0.5842,
+      "step": 1744
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.600988700564972e-05,
+      "loss": 0.4977,
+      "step": 1745
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.60075329566855e-05,
+      "loss": 0.4777,
+      "step": 1746
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.6005178907721284e-05,
+      "loss": 0.5347,
+      "step": 1747
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.600282485875707e-05,
+      "loss": 0.6622,
+      "step": 1748
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.6000470809792845e-05,
+      "loss": 0.5595,
+      "step": 1749
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.599811676082863e-05,
+      "loss": 0.5638,
+      "step": 1750
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.599576271186441e-05,
+      "loss": 0.5622,
+      "step": 1751
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.599340866290019e-05,
+      "loss": 0.5366,
+      "step": 1752
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5991054613935974e-05,
+      "loss": 0.4313,
+      "step": 1753
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.598870056497175e-05,
+      "loss": 0.4962,
+      "step": 1754
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5986346516007535e-05,
+      "loss": 0.6005,
+      "step": 1755
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.598399246704332e-05,
+      "loss": 0.6076,
+      "step": 1756
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5981638418079096e-05,
+      "loss": 0.4718,
+      "step": 1757
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.597928436911488e-05,
+      "loss": 0.6435,
+      "step": 1758
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5976930320150664e-05,
+      "loss": 0.5365,
+      "step": 1759
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.597457627118644e-05,
+      "loss": 0.4749,
+      "step": 1760
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5972222222222225e-05,
+      "loss": 0.5583,
+      "step": 1761
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.596986817325801e-05,
+      "loss": 0.573,
+      "step": 1762
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5967514124293786e-05,
+      "loss": 0.5344,
+      "step": 1763
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.596516007532957e-05,
+      "loss": 0.6334,
+      "step": 1764
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5962806026365354e-05,
+      "loss": 0.6358,
+      "step": 1765
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.596045197740113e-05,
+      "loss": 0.5728,
+      "step": 1766
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5958097928436915e-05,
+      "loss": 0.5968,
+      "step": 1767
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.59557438794727e-05,
+      "loss": 0.4984,
+      "step": 1768
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5953389830508476e-05,
+      "loss": 0.609,
+      "step": 1769
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.595103578154426e-05,
+      "loss": 0.4984,
+      "step": 1770
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.594868173258004e-05,
+      "loss": 0.6784,
+      "step": 1771
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.594632768361582e-05,
+      "loss": 0.5458,
+      "step": 1772
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5943973634651605e-05,
+      "loss": 0.569,
+      "step": 1773
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.594161958568738e-05,
+      "loss": 0.6591,
+      "step": 1774
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5939265536723166e-05,
+      "loss": 0.5717,
+      "step": 1775
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.593691148775895e-05,
+      "loss": 0.6267,
+      "step": 1776
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.593455743879473e-05,
+      "loss": 0.5508,
+      "step": 1777
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.593220338983051e-05,
+      "loss": 0.5777,
+      "step": 1778
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.5929849340866295e-05,
+      "loss": 0.4851,
+      "step": 1779
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.592749529190207e-05,
+      "loss": 0.4335,
+      "step": 1780
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.5925141242937856e-05,
+      "loss": 0.5605,
+      "step": 1781
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.592278719397364e-05,
+      "loss": 0.461,
+      "step": 1782
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.592043314500942e-05,
+      "loss": 0.5216,
+      "step": 1783
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.59180790960452e-05,
+      "loss": 0.6091,
+      "step": 1784
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.5915725047080985e-05,
+      "loss": 0.4632,
+      "step": 1785
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.591337099811676e-05,
+      "loss": 0.5109,
+      "step": 1786
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.5911016949152546e-05,
+      "loss": 0.4464,
+      "step": 1787
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.590866290018832e-05,
+      "loss": 0.5445,
+      "step": 1788
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.590630885122411e-05,
+      "loss": 0.5632,
+      "step": 1789
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.590395480225989e-05,
+      "loss": 0.5349,
+      "step": 1790
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.590160075329567e-05,
+      "loss": 0.5623,
+      "step": 1791
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.589924670433145e-05,
+      "loss": 0.4843,
+      "step": 1792
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.5896892655367236e-05,
+      "loss": 0.4473,
+      "step": 1793
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.589453860640301e-05,
+      "loss": 0.5419,
+      "step": 1794
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.58921845574388e-05,
+      "loss": 0.525,
+      "step": 1795
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.588983050847458e-05,
+      "loss": 0.5238,
+      "step": 1796
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.588747645951036e-05,
+      "loss": 0.4099,
+      "step": 1797
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.588512241054614e-05,
+      "loss": 0.3777,
+      "step": 1798
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.5882768361581926e-05,
+      "loss": 0.6051,
+      "step": 1799
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.58804143126177e-05,
+      "loss": 0.528,
+      "step": 1800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.587806026365349e-05,
+      "loss": 0.6409,
+      "step": 1801
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.587570621468927e-05,
+      "loss": 0.4732,
+      "step": 1802
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.587335216572505e-05,
+      "loss": 0.5386,
+      "step": 1803
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.587099811676083e-05,
+      "loss": 0.5124,
+      "step": 1804
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5868644067796616e-05,
+      "loss": 0.5123,
+      "step": 1805
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.586629001883239e-05,
+      "loss": 0.6335,
+      "step": 1806
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.586393596986818e-05,
+      "loss": 0.4779,
+      "step": 1807
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5861581920903954e-05,
+      "loss": 0.4725,
+      "step": 1808
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.585922787193974e-05,
+      "loss": 0.5324,
+      "step": 1809
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.585687382297552e-05,
+      "loss": 0.4714,
+      "step": 1810
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.58545197740113e-05,
+      "loss": 0.489,
+      "step": 1811
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.585216572504708e-05,
+      "loss": 0.5309,
+      "step": 1812
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.584981167608287e-05,
+      "loss": 0.4759,
+      "step": 1813
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5847457627118644e-05,
+      "loss": 0.7659,
+      "step": 1814
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.584510357815443e-05,
+      "loss": 0.5423,
+      "step": 1815
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.584274952919021e-05,
+      "loss": 0.6168,
+      "step": 1816
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.584039548022599e-05,
+      "loss": 0.4681,
+      "step": 1817
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.583804143126177e-05,
+      "loss": 0.6491,
+      "step": 1818
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.583568738229756e-05,
+      "loss": 0.5342,
+      "step": 1819
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5833333333333334e-05,
+      "loss": 0.473,
+      "step": 1820
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.583097928436912e-05,
+      "loss": 0.565,
+      "step": 1821
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.58286252354049e-05,
+      "loss": 0.5584,
+      "step": 1822
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.582627118644068e-05,
+      "loss": 0.507,
+      "step": 1823
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.582391713747646e-05,
+      "loss": 0.5233,
+      "step": 1824
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.582156308851224e-05,
+      "loss": 0.5417,
+      "step": 1825
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5819209039548024e-05,
+      "loss": 0.5781,
+      "step": 1826
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.581685499058381e-05,
+      "loss": 0.5609,
+      "step": 1827
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5814500941619585e-05,
+      "loss": 0.4724,
+      "step": 1828
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.581214689265537e-05,
+      "loss": 0.5981,
+      "step": 1829
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.580979284369115e-05,
+      "loss": 0.5593,
+      "step": 1830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.580743879472693e-05,
+      "loss": 0.5256,
+      "step": 1831
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5805084745762714e-05,
+      "loss": 0.5226,
+      "step": 1832
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.58027306967985e-05,
+      "loss": 0.5426,
+      "step": 1833
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5800376647834275e-05,
+      "loss": 0.5132,
+      "step": 1834
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.579802259887006e-05,
+      "loss": 0.5585,
+      "step": 1835
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.579566854990584e-05,
+      "loss": 0.5211,
+      "step": 1836
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.579331450094162e-05,
+      "loss": 0.555,
+      "step": 1837
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5790960451977404e-05,
+      "loss": 0.4953,
+      "step": 1838
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.578860640301319e-05,
+      "loss": 0.5224,
+      "step": 1839
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5786252354048965e-05,
+      "loss": 0.4644,
+      "step": 1840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.578389830508475e-05,
+      "loss": 0.4857,
+      "step": 1841
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5781544256120526e-05,
+      "loss": 0.6456,
+      "step": 1842
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.577919020715631e-05,
+      "loss": 0.4836,
+      "step": 1843
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5776836158192094e-05,
+      "loss": 0.5852,
+      "step": 1844
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.577448210922787e-05,
+      "loss": 0.5206,
+      "step": 1845
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5772128060263655e-05,
+      "loss": 0.5786,
+      "step": 1846
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.576977401129944e-05,
+      "loss": 0.5667,
+      "step": 1847
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5767419962335216e-05,
+      "loss": 0.507,
+      "step": 1848
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5765065913371e-05,
+      "loss": 0.5826,
+      "step": 1849
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5762711864406784e-05,
+      "loss": 0.4978,
+      "step": 1850
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.576035781544256e-05,
+      "loss": 0.6572,
+      "step": 1851
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5758003766478345e-05,
+      "loss": 0.4455,
+      "step": 1852
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.575564971751413e-05,
+      "loss": 0.5588,
+      "step": 1853
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5753295668549906e-05,
+      "loss": 0.4231,
+      "step": 1854
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.575094161958569e-05,
+      "loss": 0.4761,
+      "step": 1855
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5748587570621474e-05,
+      "loss": 0.6295,
+      "step": 1856
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.574623352165725e-05,
+      "loss": 0.5048,
+      "step": 1857
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5743879472693035e-05,
+      "loss": 0.5222,
+      "step": 1858
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.574152542372881e-05,
+      "loss": 0.4613,
+      "step": 1859
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.57391713747646e-05,
+      "loss": 0.4883,
+      "step": 1860
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.573681732580038e-05,
+      "loss": 0.479,
+      "step": 1861
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.573446327683616e-05,
+      "loss": 0.5448,
+      "step": 1862
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.573210922787194e-05,
+      "loss": 0.5835,
+      "step": 1863
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5729755178907725e-05,
+      "loss": 0.4204,
+      "step": 1864
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.57274011299435e-05,
+      "loss": 0.5746,
+      "step": 1865
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5725047080979286e-05,
+      "loss": 0.602,
+      "step": 1866
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.572269303201507e-05,
+      "loss": 0.6809,
+      "step": 1867
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.572033898305085e-05,
+      "loss": 0.5444,
+      "step": 1868
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.571798493408663e-05,
+      "loss": 0.4741,
+      "step": 1869
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5715630885122415e-05,
+      "loss": 0.3841,
+      "step": 1870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.571327683615819e-05,
+      "loss": 0.5508,
+      "step": 1871
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5710922787193976e-05,
+      "loss": 0.4713,
+      "step": 1872
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.570856873822976e-05,
+      "loss": 0.4536,
+      "step": 1873
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.570621468926554e-05,
+      "loss": 0.4478,
+      "step": 1874
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.570386064030132e-05,
+      "loss": 0.4871,
+      "step": 1875
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.57015065913371e-05,
+      "loss": 0.5409,
+      "step": 1876
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.569915254237289e-05,
+      "loss": 0.6287,
+      "step": 1877
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5696798493408666e-05,
+      "loss": 0.7449,
+      "step": 1878
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.569444444444444e-05,
+      "loss": 0.4984,
+      "step": 1879
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5692090395480234e-05,
+      "loss": 0.4605,
+      "step": 1880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.568973634651601e-05,
+      "loss": 0.5489,
+      "step": 1881
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.568738229755179e-05,
+      "loss": 0.6178,
+      "step": 1882
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.568502824858757e-05,
+      "loss": 0.4879,
+      "step": 1883
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5682674199623356e-05,
+      "loss": 0.4676,
+      "step": 1884
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.568032015065913e-05,
+      "loss": 0.6087,
+      "step": 1885
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.567796610169492e-05,
+      "loss": 0.5174,
+      "step": 1886
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.56756120527307e-05,
+      "loss": 0.5617,
+      "step": 1887
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.567325800376648e-05,
+      "loss": 0.5014,
+      "step": 1888
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.567090395480226e-05,
+      "loss": 0.4638,
+      "step": 1889
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5668549905838046e-05,
+      "loss": 0.6897,
+      "step": 1890
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.566619585687382e-05,
+      "loss": 0.6515,
+      "step": 1891
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.566384180790961e-05,
+      "loss": 0.5315,
+      "step": 1892
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5661487758945384e-05,
+      "loss": 0.5721,
+      "step": 1893
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5659133709981175e-05,
+      "loss": 0.4255,
+      "step": 1894
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.565677966101695e-05,
+      "loss": 0.5368,
+      "step": 1895
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.565442561205273e-05,
+      "loss": 0.5889,
+      "step": 1896
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.565207156308852e-05,
+      "loss": 0.6,
+      "step": 1897
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.56497175141243e-05,
+      "loss": 0.5081,
+      "step": 1898
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5647363465160074e-05,
+      "loss": 0.5107,
+      "step": 1899
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.564500941619586e-05,
+      "loss": 0.5609,
+      "step": 1900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.564265536723164e-05,
+      "loss": 0.4925,
+      "step": 1901
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.564030131826742e-05,
+      "loss": 0.5374,
+      "step": 1902
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.56379472693032e-05,
+      "loss": 0.6085,
+      "step": 1903
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.563559322033899e-05,
+      "loss": 0.4408,
+      "step": 1904
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5633239171374764e-05,
+      "loss": 0.4743,
+      "step": 1905
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.563088512241055e-05,
+      "loss": 0.6174,
+      "step": 1906
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.562853107344633e-05,
+      "loss": 0.5594,
+      "step": 1907
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5626177024482116e-05,
+      "loss": 0.5383,
+      "step": 1908
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.562382297551789e-05,
+      "loss": 0.4631,
+      "step": 1909
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.562146892655367e-05,
+      "loss": 0.4721,
+      "step": 1910
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.561911487758946e-05,
+      "loss": 0.4373,
+      "step": 1911
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.561676082862524e-05,
+      "loss": 0.6091,
+      "step": 1912
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5614406779661015e-05,
+      "loss": 0.579,
+      "step": 1913
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5612052730696806e-05,
+      "loss": 0.5497,
+      "step": 1914
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.560969868173258e-05,
+      "loss": 0.5186,
+      "step": 1915
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.560734463276836e-05,
+      "loss": 0.4943,
+      "step": 1916
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5604990583804144e-05,
+      "loss": 0.5114,
+      "step": 1917
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.560263653483993e-05,
+      "loss": 0.5721,
+      "step": 1918
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5600282485875705e-05,
+      "loss": 0.6207,
+      "step": 1919
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.559792843691149e-05,
+      "loss": 0.5621,
+      "step": 1920
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.559557438794727e-05,
+      "loss": 0.5615,
+      "step": 1921
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.559322033898305e-05,
+      "loss": 0.6848,
+      "step": 1922
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5590866290018834e-05,
+      "loss": 0.5786,
+      "step": 1923
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.558851224105462e-05,
+      "loss": 0.4635,
+      "step": 1924
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.55861581920904e-05,
+      "loss": 0.5264,
+      "step": 1925
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.558380414312618e-05,
+      "loss": 0.6074,
+      "step": 1926
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5581450094161956e-05,
+      "loss": 0.533,
+      "step": 1927
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.557909604519775e-05,
+      "loss": 0.6225,
+      "step": 1928
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5576741996233524e-05,
+      "loss": 0.6496,
+      "step": 1929
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.55743879472693e-05,
+      "loss": 0.5514,
+      "step": 1930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.557203389830509e-05,
+      "loss": 0.42,
+      "step": 1931
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.556967984934087e-05,
+      "loss": 0.5256,
+      "step": 1932
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5567325800376646e-05,
+      "loss": 0.5584,
+      "step": 1933
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.556497175141243e-05,
+      "loss": 0.5683,
+      "step": 1934
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5562617702448214e-05,
+      "loss": 0.4415,
+      "step": 1935
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.556026365348399e-05,
+      "loss": 0.5844,
+      "step": 1936
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5557909604519775e-05,
+      "loss": 0.5842,
+      "step": 1937
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.555555555555556e-05,
+      "loss": 0.465,
+      "step": 1938
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.555320150659134e-05,
+      "loss": 0.4962,
+      "step": 1939
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.555084745762712e-05,
+      "loss": 0.5059,
+      "step": 1940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5548493408662904e-05,
+      "loss": 0.5529,
+      "step": 1941
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.554613935969869e-05,
+      "loss": 0.6004,
+      "step": 1942
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5543785310734465e-05,
+      "loss": 0.4674,
+      "step": 1943
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.554143126177024e-05,
+      "loss": 0.4757,
+      "step": 1944
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.553907721280603e-05,
+      "loss": 0.591,
+      "step": 1945
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.553672316384181e-05,
+      "loss": 0.531,
+      "step": 1946
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.553436911487759e-05,
+      "loss": 0.4563,
+      "step": 1947
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.553201506591338e-05,
+      "loss": 0.4935,
+      "step": 1948
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5529661016949155e-05,
+      "loss": 0.4833,
+      "step": 1949
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.552730696798493e-05,
+      "loss": 0.4974,
+      "step": 1950
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5524952919020716e-05,
+      "loss": 0.4191,
+      "step": 1951
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.55225988700565e-05,
+      "loss": 0.4672,
+      "step": 1952
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.552024482109228e-05,
+      "loss": 0.5292,
+      "step": 1953
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.551789077212806e-05,
+      "loss": 0.4765,
+      "step": 1954
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5515536723163845e-05,
+      "loss": 0.5042,
+      "step": 1955
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.551318267419963e-05,
+      "loss": 0.4545,
+      "step": 1956
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5510828625235406e-05,
+      "loss": 0.4725,
+      "step": 1957
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.550847457627119e-05,
+      "loss": 0.4512,
+      "step": 1958
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5506120527306974e-05,
+      "loss": 0.456,
+      "step": 1959
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.550376647834275e-05,
+      "loss": 0.4093,
+      "step": 1960
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.550141242937853e-05,
+      "loss": 0.3909,
+      "step": 1961
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.549905838041432e-05,
+      "loss": 0.4886,
+      "step": 1962
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5496704331450096e-05,
+      "loss": 0.5867,
+      "step": 1963
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.549435028248587e-05,
+      "loss": 0.4676,
+      "step": 1964
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5491996233521664e-05,
+      "loss": 0.6013,
+      "step": 1965
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.548964218455744e-05,
+      "loss": 0.4536,
+      "step": 1966
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.548728813559322e-05,
+      "loss": 0.4021,
+      "step": 1967
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5484934086629e-05,
+      "loss": 0.6665,
+      "step": 1968
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5482580037664786e-05,
+      "loss": 0.6261,
+      "step": 1969
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.548022598870056e-05,
+      "loss": 0.5359,
+      "step": 1970
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.547787193973635e-05,
+      "loss": 0.5207,
+      "step": 1971
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.547551789077213e-05,
+      "loss": 0.4952,
+      "step": 1972
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5473163841807915e-05,
+      "loss": 0.4852,
+      "step": 1973
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.547080979284369e-05,
+      "loss": 0.5115,
+      "step": 1974
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5468455743879476e-05,
+      "loss": 0.505,
+      "step": 1975
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.546610169491526e-05,
+      "loss": 0.4887,
+      "step": 1976
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.546374764595104e-05,
+      "loss": 0.5663,
+      "step": 1977
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5461393596986814e-05,
+      "loss": 0.4554,
+      "step": 1978
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5459039548022605e-05,
+      "loss": 0.5218,
+      "step": 1979
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.545668549905838e-05,
+      "loss": 0.5417,
+      "step": 1980
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.545433145009416e-05,
+      "loss": 0.4331,
+      "step": 1981
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.545197740112995e-05,
+      "loss": 0.626,
+      "step": 1982
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.544962335216573e-05,
+      "loss": 0.4845,
+      "step": 1983
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5447269303201504e-05,
+      "loss": 0.5004,
+      "step": 1984
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.544491525423729e-05,
+      "loss": 0.5016,
+      "step": 1985
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.544256120527307e-05,
+      "loss": 0.5263,
+      "step": 1986
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5440207156308856e-05,
+      "loss": 0.5312,
+      "step": 1987
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.543785310734463e-05,
+      "loss": 0.4283,
+      "step": 1988
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.543549905838042e-05,
+      "loss": 0.4779,
+      "step": 1989
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.54331450094162e-05,
+      "loss": 0.5588,
+      "step": 1990
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.543079096045198e-05,
+      "loss": 0.6557,
+      "step": 1991
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.542843691148776e-05,
+      "loss": 0.6026,
+      "step": 1992
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5426082862523546e-05,
+      "loss": 0.5201,
+      "step": 1993
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.542372881355932e-05,
+      "loss": 0.5253,
+      "step": 1994
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.54213747645951e-05,
+      "loss": 0.4726,
+      "step": 1995
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.541902071563089e-05,
+      "loss": 0.6189,
+      "step": 1996
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.541666666666667e-05,
+      "loss": 0.5074,
+      "step": 1997
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5414312617702445e-05,
+      "loss": 0.4144,
+      "step": 1998
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5411958568738236e-05,
+      "loss": 0.484,
+      "step": 1999
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.540960451977401e-05,
+      "loss": 0.4458,
+      "step": 2000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.540725047080979e-05,
+      "loss": 0.4966,
+      "step": 2001
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5404896421845574e-05,
+      "loss": 0.5718,
+      "step": 2002
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.540254237288136e-05,
+      "loss": 0.416,
+      "step": 2003
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.540018832391714e-05,
+      "loss": 0.4647,
+      "step": 2004
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.539783427495292e-05,
+      "loss": 0.5042,
+      "step": 2005
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.53954802259887e-05,
+      "loss": 0.5624,
+      "step": 2006
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.539312617702449e-05,
+      "loss": 0.4273,
+      "step": 2007
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5390772128060264e-05,
+      "loss": 0.5558,
+      "step": 2008
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.538841807909605e-05,
+      "loss": 0.5369,
+      "step": 2009
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.538606403013183e-05,
+      "loss": 0.5198,
+      "step": 2010
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.538370998116761e-05,
+      "loss": 0.6098,
+      "step": 2011
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5381355932203387e-05,
+      "loss": 0.4313,
+      "step": 2012
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.537900188323918e-05,
+      "loss": 0.4835,
+      "step": 2013
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5376647834274954e-05,
+      "loss": 0.4478,
+      "step": 2014
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.537429378531073e-05,
+      "loss": 0.5029,
+      "step": 2015
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.537193973634652e-05,
+      "loss": 0.5363,
+      "step": 2016
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.53695856873823e-05,
+      "loss": 0.4664,
+      "step": 2017
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.536723163841808e-05,
+      "loss": 0.6602,
+      "step": 2018
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.536487758945386e-05,
+      "loss": 0.578,
+      "step": 2019
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5362523540489644e-05,
+      "loss": 0.6891,
+      "step": 2020
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.536016949152543e-05,
+      "loss": 0.4884,
+      "step": 2021
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5357815442561205e-05,
+      "loss": 0.5316,
+      "step": 2022
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.535546139359699e-05,
+      "loss": 0.6828,
+      "step": 2023
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.535310734463277e-05,
+      "loss": 0.4627,
+      "step": 2024
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.535075329566855e-05,
+      "loss": 0.4302,
+      "step": 2025
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5348399246704334e-05,
+      "loss": 0.4515,
+      "step": 2026
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.534604519774012e-05,
+      "loss": 0.6073,
+      "step": 2027
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5343691148775895e-05,
+      "loss": 0.5593,
+      "step": 2028
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.534133709981167e-05,
+      "loss": 0.5023,
+      "step": 2029
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.533898305084746e-05,
+      "loss": 0.5828,
+      "step": 2030
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.533662900188324e-05,
+      "loss": 0.5086,
+      "step": 2031
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.533427495291902e-05,
+      "loss": 0.5007,
+      "step": 2032
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.533192090395481e-05,
+      "loss": 0.5566,
+      "step": 2033
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5329566854990585e-05,
+      "loss": 0.5712,
+      "step": 2034
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.532721280602637e-05,
+      "loss": 0.578,
+      "step": 2035
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5324858757062147e-05,
+      "loss": 0.5718,
+      "step": 2036
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.532250470809793e-05,
+      "loss": 0.665,
+      "step": 2037
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5320150659133714e-05,
+      "loss": 0.5046,
+      "step": 2038
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.531779661016949e-05,
+      "loss": 0.5559,
+      "step": 2039
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5315442561205275e-05,
+      "loss": 0.4482,
+      "step": 2040
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.531308851224106e-05,
+      "loss": 0.5763,
+      "step": 2041
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5310734463276837e-05,
+      "loss": 0.5886,
+      "step": 2042
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.530838041431262e-05,
+      "loss": 0.6628,
+      "step": 2043
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5306026365348404e-05,
+      "loss": 0.4722,
+      "step": 2044
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.530367231638418e-05,
+      "loss": 0.486,
+      "step": 2045
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5301318267419965e-05,
+      "loss": 0.4965,
+      "step": 2046
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.529896421845575e-05,
+      "loss": 0.5371,
+      "step": 2047
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5296610169491527e-05,
+      "loss": 0.6226,
+      "step": 2048
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5294256120527304e-05,
+      "loss": 0.5445,
+      "step": 2049
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5291902071563094e-05,
+      "loss": 0.4366,
+      "step": 2050
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.528954802259887e-05,
+      "loss": 0.6041,
+      "step": 2051
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5287193973634655e-05,
+      "loss": 0.5117,
+      "step": 2052
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.528483992467043e-05,
+      "loss": 0.5056,
+      "step": 2053
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5282485875706217e-05,
+      "loss": 0.5244,
+      "step": 2054
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5280131826742e-05,
+      "loss": 0.552,
+      "step": 2055
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.527777777777778e-05,
+      "loss": 0.4876,
+      "step": 2056
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.527542372881356e-05,
+      "loss": 0.5854,
+      "step": 2057
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5273069679849345e-05,
+      "loss": 0.4941,
+      "step": 2058
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.527071563088512e-05,
+      "loss": 0.6248,
+      "step": 2059
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5268361581920906e-05,
+      "loss": 0.535,
+      "step": 2060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.526600753295669e-05,
+      "loss": 0.51,
+      "step": 2061
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.526365348399247e-05,
+      "loss": 0.4018,
+      "step": 2062
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.526129943502825e-05,
+      "loss": 0.587,
+      "step": 2063
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5258945386064035e-05,
+      "loss": 0.5394,
+      "step": 2064
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.525659133709981e-05,
+      "loss": 0.4587,
+      "step": 2065
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5254237288135596e-05,
+      "loss": 0.4922,
+      "step": 2066
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.525188323917138e-05,
+      "loss": 0.466,
+      "step": 2067
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.524952919020716e-05,
+      "loss": 0.4787,
+      "step": 2068
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.524717514124294e-05,
+      "loss": 0.48,
+      "step": 2069
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.524482109227872e-05,
+      "loss": 0.4885,
+      "step": 2070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.52424670433145e-05,
+      "loss": 0.636,
+      "step": 2071
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5240112994350286e-05,
+      "loss": 0.5474,
+      "step": 2072
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5237758945386064e-05,
+      "loss": 0.6022,
+      "step": 2073
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.523540489642185e-05,
+      "loss": 0.3503,
+      "step": 2074
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.523305084745763e-05,
+      "loss": 0.52,
+      "step": 2075
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.523069679849341e-05,
+      "loss": 0.7183,
+      "step": 2076
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.522834274952919e-05,
+      "loss": 0.5031,
+      "step": 2077
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5225988700564976e-05,
+      "loss": 0.525,
+      "step": 2078
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5223634651600754e-05,
+      "loss": 0.426,
+      "step": 2079
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.522128060263654e-05,
+      "loss": 0.4216,
+      "step": 2080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.521892655367232e-05,
+      "loss": 0.5084,
+      "step": 2081
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.52165725047081e-05,
+      "loss": 0.6203,
+      "step": 2082
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.521421845574388e-05,
+      "loss": 0.6199,
+      "step": 2083
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5211864406779666e-05,
+      "loss": 0.57,
+      "step": 2084
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5209510357815444e-05,
+      "loss": 0.4302,
+      "step": 2085
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.520715630885123e-05,
+      "loss": 0.4752,
+      "step": 2086
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5204802259887005e-05,
+      "loss": 0.5117,
+      "step": 2087
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.520244821092279e-05,
+      "loss": 0.5092,
+      "step": 2088
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.520009416195857e-05,
+      "loss": 0.5619,
+      "step": 2089
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.519774011299435e-05,
+      "loss": 0.6229,
+      "step": 2090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5195386064030134e-05,
+      "loss": 0.4522,
+      "step": 2091
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.519303201506592e-05,
+      "loss": 0.6451,
+      "step": 2092
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5190677966101695e-05,
+      "loss": 0.5271,
+      "step": 2093
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.518832391713748e-05,
+      "loss": 0.5021,
+      "step": 2094
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.518596986817326e-05,
+      "loss": 0.4549,
+      "step": 2095
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.518361581920904e-05,
+      "loss": 0.5435,
+      "step": 2096
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5181261770244824e-05,
+      "loss": 0.4428,
+      "step": 2097
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.517890772128061e-05,
+      "loss": 0.4882,
+      "step": 2098
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5176553672316385e-05,
+      "loss": 0.5213,
+      "step": 2099
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.517419962335217e-05,
+      "loss": 0.4972,
+      "step": 2100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.517184557438795e-05,
+      "loss": 0.5481,
+      "step": 2101
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.516949152542373e-05,
+      "loss": 0.4411,
+      "step": 2102
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5167137476459514e-05,
+      "loss": 0.5125,
+      "step": 2103
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.516478342749529e-05,
+      "loss": 0.5153,
+      "step": 2104
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5162429378531075e-05,
+      "loss": 0.4678,
+      "step": 2105
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.516007532956686e-05,
+      "loss": 0.4839,
+      "step": 2106
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5157721280602636e-05,
+      "loss": 0.5742,
+      "step": 2107
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.515536723163842e-05,
+      "loss": 0.5021,
+      "step": 2108
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5153013182674204e-05,
+      "loss": 0.4891,
+      "step": 2109
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.515065913370998e-05,
+      "loss": 0.5039,
+      "step": 2110
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5148305084745765e-05,
+      "loss": 0.433,
+      "step": 2111
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.514595103578155e-05,
+      "loss": 0.59,
+      "step": 2112
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5143596986817326e-05,
+      "loss": 0.5469,
+      "step": 2113
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.514124293785311e-05,
+      "loss": 0.5549,
+      "step": 2114
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5138888888888894e-05,
+      "loss": 0.4061,
+      "step": 2115
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.513653483992467e-05,
+      "loss": 0.5515,
+      "step": 2116
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5134180790960455e-05,
+      "loss": 0.4851,
+      "step": 2117
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.513182674199624e-05,
+      "loss": 0.453,
+      "step": 2118
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5129472693032016e-05,
+      "loss": 0.4965,
+      "step": 2119
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.51271186440678e-05,
+      "loss": 0.3845,
+      "step": 2120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5124764595103584e-05,
+      "loss": 0.5538,
+      "step": 2121
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.512241054613936e-05,
+      "loss": 0.5994,
+      "step": 2122
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5120056497175145e-05,
+      "loss": 0.515,
+      "step": 2123
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.511770244821092e-05,
+      "loss": 0.4969,
+      "step": 2124
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5115348399246706e-05,
+      "loss": 0.5056,
+      "step": 2125
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.511299435028249e-05,
+      "loss": 0.4451,
+      "step": 2126
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.511064030131827e-05,
+      "loss": 0.4218,
+      "step": 2127
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.510828625235405e-05,
+      "loss": 0.569,
+      "step": 2128
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5105932203389835e-05,
+      "loss": 0.453,
+      "step": 2129
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.7692714298299709,
+      "eval_loss": 0.48256269097328186,
+      "eval_runtime": 23.104,
+      "eval_samples_per_second": 1473.902,
+      "eval_steps_per_second": 23.07,
+      "step": 2129
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.510357815442561e-05,
+      "loss": 0.5826,
+      "step": 2130
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5101224105461396e-05,
+      "loss": 0.4543,
+      "step": 2131
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.509887005649718e-05,
+      "loss": 0.4577,
+      "step": 2132
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.509651600753296e-05,
+      "loss": 0.4586,
+      "step": 2133
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.509416195856874e-05,
+      "loss": 0.3438,
+      "step": 2134
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5091807909604525e-05,
+      "loss": 0.4342,
+      "step": 2135
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.50894538606403e-05,
+      "loss": 0.3943,
+      "step": 2136
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5087099811676086e-05,
+      "loss": 0.4815,
+      "step": 2137
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.508474576271187e-05,
+      "loss": 0.3903,
+      "step": 2138
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.508239171374765e-05,
+      "loss": 0.5026,
+      "step": 2139
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.508003766478343e-05,
+      "loss": 0.3979,
+      "step": 2140
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.507768361581921e-05,
+      "loss": 0.4111,
+      "step": 2141
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.507532956685499e-05,
+      "loss": 0.3912,
+      "step": 2142
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5072975517890776e-05,
+      "loss": 0.4245,
+      "step": 2143
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.507062146892655e-05,
+      "loss": 0.416,
+      "step": 2144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.506826741996234e-05,
+      "loss": 0.3512,
+      "step": 2145
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.506591337099812e-05,
+      "loss": 0.3775,
+      "step": 2146
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.50635593220339e-05,
+      "loss": 0.4773,
+      "step": 2147
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.506120527306968e-05,
+      "loss": 0.4094,
+      "step": 2148
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5058851224105466e-05,
+      "loss": 0.5193,
+      "step": 2149
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.505649717514124e-05,
+      "loss": 0.5734,
+      "step": 2150
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.505414312617703e-05,
+      "loss": 0.4035,
+      "step": 2151
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.505178907721281e-05,
+      "loss": 0.4633,
+      "step": 2152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.504943502824859e-05,
+      "loss": 0.3238,
+      "step": 2153
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.504708097928437e-05,
+      "loss": 0.3561,
+      "step": 2154
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5044726930320156e-05,
+      "loss": 0.4366,
+      "step": 2155
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.504237288135593e-05,
+      "loss": 0.4711,
+      "step": 2156
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.504001883239172e-05,
+      "loss": 0.4084,
+      "step": 2157
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5037664783427494e-05,
+      "loss": 0.3337,
+      "step": 2158
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.503531073446328e-05,
+      "loss": 0.3021,
+      "step": 2159
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.503295668549906e-05,
+      "loss": 0.3732,
+      "step": 2160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.503060263653484e-05,
+      "loss": 0.4668,
+      "step": 2161
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.502824858757062e-05,
+      "loss": 0.5012,
+      "step": 2162
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.502589453860641e-05,
+      "loss": 0.3236,
+      "step": 2163
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5023540489642184e-05,
+      "loss": 0.4898,
+      "step": 2164
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.502118644067797e-05,
+      "loss": 0.4922,
+      "step": 2165
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.501883239171375e-05,
+      "loss": 0.3428,
+      "step": 2166
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.501647834274953e-05,
+      "loss": 0.4079,
+      "step": 2167
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.501412429378531e-05,
+      "loss": 0.329,
+      "step": 2168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.50117702448211e-05,
+      "loss": 0.3665,
+      "step": 2169
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5009416195856874e-05,
+      "loss": 0.5566,
+      "step": 2170
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.500706214689266e-05,
+      "loss": 0.4378,
+      "step": 2171
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.500470809792844e-05,
+      "loss": 0.3564,
+      "step": 2172
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.500235404896422e-05,
+      "loss": 0.4779,
+      "step": 2173
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5e-05,
+      "loss": 0.5447,
+      "step": 2174
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.499764595103578e-05,
+      "loss": 0.3601,
+      "step": 2175
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.499529190207157e-05,
+      "loss": 0.4555,
+      "step": 2176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.499293785310735e-05,
+      "loss": 0.3108,
+      "step": 2177
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4990583804143125e-05,
+      "loss": 0.4777,
+      "step": 2178
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.498822975517891e-05,
+      "loss": 0.3227,
+      "step": 2179
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.498587570621469e-05,
+      "loss": 0.46,
+      "step": 2180
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.498352165725047e-05,
+      "loss": 0.271,
+      "step": 2181
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4981167608286254e-05,
+      "loss": 0.4699,
+      "step": 2182
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.497881355932204e-05,
+      "loss": 0.3364,
+      "step": 2183
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4976459510357815e-05,
+      "loss": 0.4911,
+      "step": 2184
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.49741054613936e-05,
+      "loss": 0.5416,
+      "step": 2185
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.497175141242938e-05,
+      "loss": 0.4678,
+      "step": 2186
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.496939736346516e-05,
+      "loss": 0.3486,
+      "step": 2187
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4967043314500944e-05,
+      "loss": 0.4157,
+      "step": 2188
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.496468926553673e-05,
+      "loss": 0.3899,
+      "step": 2189
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4962335216572505e-05,
+      "loss": 0.4132,
+      "step": 2190
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.495998116760829e-05,
+      "loss": 0.4897,
+      "step": 2191
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4957627118644066e-05,
+      "loss": 0.4184,
+      "step": 2192
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.495527306967986e-05,
+      "loss": 0.4542,
+      "step": 2193
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4952919020715634e-05,
+      "loss": 0.4383,
+      "step": 2194
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.495056497175141e-05,
+      "loss": 0.2769,
+      "step": 2195
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.49482109227872e-05,
+      "loss": 0.3544,
+      "step": 2196
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.494585687382298e-05,
+      "loss": 0.3046,
+      "step": 2197
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4943502824858756e-05,
+      "loss": 0.4755,
+      "step": 2198
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.494114877589454e-05,
+      "loss": 0.3866,
+      "step": 2199
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4938794726930324e-05,
+      "loss": 0.3913,
+      "step": 2200
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.49364406779661e-05,
+      "loss": 0.3526,
+      "step": 2201
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4934086629001885e-05,
+      "loss": 0.4299,
+      "step": 2202
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.493173258003767e-05,
+      "loss": 0.3865,
+      "step": 2203
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4929378531073446e-05,
+      "loss": 0.3614,
+      "step": 2204
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.492702448210923e-05,
+      "loss": 0.4417,
+      "step": 2205
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4924670433145014e-05,
+      "loss": 0.3917,
+      "step": 2206
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.492231638418079e-05,
+      "loss": 0.4419,
+      "step": 2207
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4919962335216575e-05,
+      "loss": 0.2903,
+      "step": 2208
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.491760828625235e-05,
+      "loss": 0.3632,
+      "step": 2209
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.491525423728814e-05,
+      "loss": 0.4353,
+      "step": 2210
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.491290018832392e-05,
+      "loss": 0.4589,
+      "step": 2211
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.49105461393597e-05,
+      "loss": 0.2857,
+      "step": 2212
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.490819209039549e-05,
+      "loss": 0.3801,
+      "step": 2213
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4905838041431265e-05,
+      "loss": 0.2525,
+      "step": 2214
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.490348399246704e-05,
+      "loss": 0.4163,
+      "step": 2215
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4901129943502826e-05,
+      "loss": 0.5561,
+      "step": 2216
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.489877589453861e-05,
+      "loss": 0.4344,
+      "step": 2217
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.489642184557439e-05,
+      "loss": 0.4921,
+      "step": 2218
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.489406779661017e-05,
+      "loss": 0.5735,
+      "step": 2219
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4891713747645955e-05,
+      "loss": 0.369,
+      "step": 2220
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.488935969868173e-05,
+      "loss": 0.4306,
+      "step": 2221
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4887005649717516e-05,
+      "loss": 0.4158,
+      "step": 2222
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.48846516007533e-05,
+      "loss": 0.4718,
+      "step": 2223
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4882297551789084e-05,
+      "loss": 0.3204,
+      "step": 2224
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.487994350282486e-05,
+      "loss": 0.3638,
+      "step": 2225
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.487758945386064e-05,
+      "loss": 0.3554,
+      "step": 2226
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.487523540489643e-05,
+      "loss": 0.3619,
+      "step": 2227
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.4872881355932206e-05,
+      "loss": 0.3658,
+      "step": 2228
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.487052730696798e-05,
+      "loss": 0.3726,
+      "step": 2229
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.4868173258003774e-05,
+      "loss": 0.3754,
+      "step": 2230
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.486581920903955e-05,
+      "loss": 0.4727,
+      "step": 2231
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.486346516007533e-05,
+      "loss": 0.4763,
+      "step": 2232
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.486111111111111e-05,
+      "loss": 0.294,
+      "step": 2233
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.4858757062146896e-05,
+      "loss": 0.5063,
+      "step": 2234
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.485640301318267e-05,
+      "loss": 0.4684,
+      "step": 2235
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.485404896421846e-05,
+      "loss": 0.5026,
+      "step": 2236
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.485169491525424e-05,
+      "loss": 0.2644,
+      "step": 2237
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.484934086629002e-05,
+      "loss": 0.4787,
+      "step": 2238
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.48469868173258e-05,
+      "loss": 0.4466,
+      "step": 2239
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.4844632768361586e-05,
+      "loss": 0.3801,
+      "step": 2240
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.484227871939737e-05,
+      "loss": 0.3906,
+      "step": 2241
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.483992467043315e-05,
+      "loss": 0.3427,
+      "step": 2242
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.4837570621468924e-05,
+      "loss": 0.4057,
+      "step": 2243
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.4835216572504715e-05,
+      "loss": 0.3435,
+      "step": 2244
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.483286252354049e-05,
+      "loss": 0.3852,
+      "step": 2245
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.483050847457627e-05,
+      "loss": 0.3455,
+      "step": 2246
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.482815442561206e-05,
+      "loss": 0.3675,
+      "step": 2247
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.482580037664784e-05,
+      "loss": 0.4146,
+      "step": 2248
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4823446327683614e-05,
+      "loss": 0.3326,
+      "step": 2249
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.48210922787194e-05,
+      "loss": 0.3578,
+      "step": 2250
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.481873822975518e-05,
+      "loss": 0.3375,
+      "step": 2251
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.481638418079096e-05,
+      "loss": 0.6612,
+      "step": 2252
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.481403013182674e-05,
+      "loss": 0.4411,
+      "step": 2253
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.481167608286253e-05,
+      "loss": 0.3547,
+      "step": 2254
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.480932203389831e-05,
+      "loss": 0.2948,
+      "step": 2255
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.480696798493409e-05,
+      "loss": 0.3389,
+      "step": 2256
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.480461393596987e-05,
+      "loss": 0.3237,
+      "step": 2257
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4802259887005656e-05,
+      "loss": 0.3854,
+      "step": 2258
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.479990583804143e-05,
+      "loss": 0.4024,
+      "step": 2259
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.479755178907721e-05,
+      "loss": 0.4036,
+      "step": 2260
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4795197740113e-05,
+      "loss": 0.5163,
+      "step": 2261
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.479284369114878e-05,
+      "loss": 0.3311,
+      "step": 2262
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4790489642184555e-05,
+      "loss": 0.3554,
+      "step": 2263
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4788135593220346e-05,
+      "loss": 0.4476,
+      "step": 2264
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.478578154425612e-05,
+      "loss": 0.2699,
+      "step": 2265
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.47834274952919e-05,
+      "loss": 0.3356,
+      "step": 2266
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4781073446327684e-05,
+      "loss": 0.4194,
+      "step": 2267
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.477871939736347e-05,
+      "loss": 0.4427,
+      "step": 2268
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4776365348399245e-05,
+      "loss": 0.3943,
+      "step": 2269
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.477401129943503e-05,
+      "loss": 0.5866,
+      "step": 2270
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.477165725047081e-05,
+      "loss": 0.4156,
+      "step": 2271
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.47693032015066e-05,
+      "loss": 0.4239,
+      "step": 2272
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4766949152542374e-05,
+      "loss": 0.4312,
+      "step": 2273
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.476459510357816e-05,
+      "loss": 0.4215,
+      "step": 2274
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.476224105461394e-05,
+      "loss": 0.3943,
+      "step": 2275
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.475988700564972e-05,
+      "loss": 0.4238,
+      "step": 2276
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4757532956685496e-05,
+      "loss": 0.3821,
+      "step": 2277
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.475517890772129e-05,
+      "loss": 0.4069,
+      "step": 2278
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4752824858757064e-05,
+      "loss": 0.5432,
+      "step": 2279
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.475047080979284e-05,
+      "loss": 0.3796,
+      "step": 2280
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.474811676082863e-05,
+      "loss": 0.4567,
+      "step": 2281
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.474576271186441e-05,
+      "loss": 0.4138,
+      "step": 2282
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4743408662900186e-05,
+      "loss": 0.4334,
+      "step": 2283
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.474105461393597e-05,
+      "loss": 0.34,
+      "step": 2284
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4738700564971754e-05,
+      "loss": 0.5009,
+      "step": 2285
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.473634651600753e-05,
+      "loss": 0.4199,
+      "step": 2286
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4733992467043315e-05,
+      "loss": 0.3497,
+      "step": 2287
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.47316384180791e-05,
+      "loss": 0.4368,
+      "step": 2288
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.472928436911488e-05,
+      "loss": 0.3431,
+      "step": 2289
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.472693032015066e-05,
+      "loss": 0.3833,
+      "step": 2290
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4724576271186444e-05,
+      "loss": 0.4254,
+      "step": 2291
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.472222222222223e-05,
+      "loss": 0.2665,
+      "step": 2292
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4719868173258005e-05,
+      "loss": 0.4305,
+      "step": 2293
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.471751412429378e-05,
+      "loss": 0.3942,
+      "step": 2294
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.471516007532957e-05,
+      "loss": 0.539,
+      "step": 2295
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.471280602636535e-05,
+      "loss": 0.3592,
+      "step": 2296
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.471045197740113e-05,
+      "loss": 0.4309,
+      "step": 2297
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.470809792843692e-05,
+      "loss": 0.4302,
+      "step": 2298
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4705743879472695e-05,
+      "loss": 0.3685,
+      "step": 2299
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.470338983050847e-05,
+      "loss": 0.2704,
+      "step": 2300
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4701035781544256e-05,
+      "loss": 0.4266,
+      "step": 2301
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.469868173258004e-05,
+      "loss": 0.4691,
+      "step": 2302
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4696327683615824e-05,
+      "loss": 0.3019,
+      "step": 2303
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.46939736346516e-05,
+      "loss": 0.39,
+      "step": 2304
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4691619585687385e-05,
+      "loss": 0.4118,
+      "step": 2305
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.468926553672317e-05,
+      "loss": 0.3876,
+      "step": 2306
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4686911487758946e-05,
+      "loss": 0.5173,
+      "step": 2307
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.468455743879473e-05,
+      "loss": 0.5315,
+      "step": 2308
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4682203389830514e-05,
+      "loss": 0.4454,
+      "step": 2309
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.467984934086629e-05,
+      "loss": 0.418,
+      "step": 2310
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.467749529190207e-05,
+      "loss": 0.3873,
+      "step": 2311
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.467514124293786e-05,
+      "loss": 0.4443,
+      "step": 2312
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.4672787193973636e-05,
+      "loss": 0.3838,
+      "step": 2313
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.4670433145009413e-05,
+      "loss": 0.3304,
+      "step": 2314
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.4668079096045204e-05,
+      "loss": 0.3948,
+      "step": 2315
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.466572504708098e-05,
+      "loss": 0.4018,
+      "step": 2316
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.466337099811676e-05,
+      "loss": 0.4434,
+      "step": 2317
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.466101694915254e-05,
+      "loss": 0.3805,
+      "step": 2318
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.4658662900188326e-05,
+      "loss": 0.4395,
+      "step": 2319
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.465630885122411e-05,
+      "loss": 0.3402,
+      "step": 2320
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.465395480225989e-05,
+      "loss": 0.445,
+      "step": 2321
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.465160075329567e-05,
+      "loss": 0.4563,
+      "step": 2322
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.4649246704331455e-05,
+      "loss": 0.3507,
+      "step": 2323
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.464689265536723e-05,
+      "loss": 0.3348,
+      "step": 2324
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.4644538606403016e-05,
+      "loss": 0.5451,
+      "step": 2325
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.46421845574388e-05,
+      "loss": 0.3423,
+      "step": 2326
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.463983050847458e-05,
+      "loss": 0.4681,
+      "step": 2327
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.4637476459510355e-05,
+      "loss": 0.3566,
+      "step": 2328
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.4635122410546145e-05,
+      "loss": 0.3408,
+      "step": 2329
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.463276836158192e-05,
+      "loss": 0.4714,
+      "step": 2330
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.46304143126177e-05,
+      "loss": 0.3679,
+      "step": 2331
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.462806026365349e-05,
+      "loss": 0.302,
+      "step": 2332
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.462570621468927e-05,
+      "loss": 0.4951,
+      "step": 2333
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.462335216572505e-05,
+      "loss": 0.3243,
+      "step": 2334
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.462099811676083e-05,
+      "loss": 0.4063,
+      "step": 2335
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.461864406779661e-05,
+      "loss": 0.4899,
+      "step": 2336
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.4616290018832396e-05,
+      "loss": 0.5007,
+      "step": 2337
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.4613935969868173e-05,
+      "loss": 0.3407,
+      "step": 2338
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.461158192090396e-05,
+      "loss": 0.4984,
+      "step": 2339
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.460922787193974e-05,
+      "loss": 0.4042,
+      "step": 2340
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.460687382297552e-05,
+      "loss": 0.3517,
+      "step": 2341
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.46045197740113e-05,
+      "loss": 0.2831,
+      "step": 2342
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.4602165725047086e-05,
+      "loss": 0.4884,
+      "step": 2343
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.4599811676082863e-05,
+      "loss": 0.2459,
+      "step": 2344
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.459745762711864e-05,
+      "loss": 0.401,
+      "step": 2345
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.459510357815443e-05,
+      "loss": 0.4205,
+      "step": 2346
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.459274952919021e-05,
+      "loss": 0.3397,
+      "step": 2347
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.4590395480225986e-05,
+      "loss": 0.4352,
+      "step": 2348
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.4588041431261776e-05,
+      "loss": 0.4169,
+      "step": 2349
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.4585687382297553e-05,
+      "loss": 0.4515,
+      "step": 2350
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.458333333333334e-05,
+      "loss": 0.458,
+      "step": 2351
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.4580979284369114e-05,
+      "loss": 0.4275,
+      "step": 2352
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.45786252354049e-05,
+      "loss": 0.3808,
+      "step": 2353
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.457627118644068e-05,
+      "loss": 0.2753,
+      "step": 2354
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.457391713747646e-05,
+      "loss": 0.413,
+      "step": 2355
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4571563088512243e-05,
+      "loss": 0.4837,
+      "step": 2356
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.456920903954803e-05,
+      "loss": 0.4081,
+      "step": 2357
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4566854990583804e-05,
+      "loss": 0.3041,
+      "step": 2358
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.456450094161959e-05,
+      "loss": 0.481,
+      "step": 2359
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.456214689265537e-05,
+      "loss": 0.2671,
+      "step": 2360
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.455979284369115e-05,
+      "loss": 0.3126,
+      "step": 2361
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4557438794726933e-05,
+      "loss": 0.4714,
+      "step": 2362
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.455508474576272e-05,
+      "loss": 0.2895,
+      "step": 2363
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4552730696798494e-05,
+      "loss": 0.3553,
+      "step": 2364
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.455037664783427e-05,
+      "loss": 0.2952,
+      "step": 2365
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.454802259887006e-05,
+      "loss": 0.4427,
+      "step": 2366
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.454566854990584e-05,
+      "loss": 0.4471,
+      "step": 2367
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4543314500941623e-05,
+      "loss": 0.4843,
+      "step": 2368
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.45409604519774e-05,
+      "loss": 0.3733,
+      "step": 2369
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4538606403013184e-05,
+      "loss": 0.4941,
+      "step": 2370
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.453625235404897e-05,
+      "loss": 0.4262,
+      "step": 2371
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4533898305084746e-05,
+      "loss": 0.5048,
+      "step": 2372
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.453154425612053e-05,
+      "loss": 0.5404,
+      "step": 2373
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.452919020715631e-05,
+      "loss": 0.3465,
+      "step": 2374
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.452683615819209e-05,
+      "loss": 0.2917,
+      "step": 2375
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4524482109227874e-05,
+      "loss": 0.4748,
+      "step": 2376
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.452212806026366e-05,
+      "loss": 0.5302,
+      "step": 2377
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4519774011299436e-05,
+      "loss": 0.4079,
+      "step": 2378
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.451741996233522e-05,
+      "loss": 0.2451,
+      "step": 2379
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4515065913371e-05,
+      "loss": 0.405,
+      "step": 2380
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.451271186440678e-05,
+      "loss": 0.2847,
+      "step": 2381
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4510357815442564e-05,
+      "loss": 0.5197,
+      "step": 2382
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.450800376647835e-05,
+      "loss": 0.4251,
+      "step": 2383
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4505649717514126e-05,
+      "loss": 0.4104,
+      "step": 2384
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.450329566854991e-05,
+      "loss": 0.5035,
+      "step": 2385
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4500941619585687e-05,
+      "loss": 0.455,
+      "step": 2386
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.449858757062147e-05,
+      "loss": 0.467,
+      "step": 2387
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4496233521657254e-05,
+      "loss": 0.5135,
+      "step": 2388
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.449387947269303e-05,
+      "loss": 0.4125,
+      "step": 2389
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4491525423728816e-05,
+      "loss": 0.7081,
+      "step": 2390
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.44891713747646e-05,
+      "loss": 0.3717,
+      "step": 2391
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4486817325800377e-05,
+      "loss": 0.4643,
+      "step": 2392
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.448446327683616e-05,
+      "loss": 0.3344,
+      "step": 2393
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4482109227871944e-05,
+      "loss": 0.4275,
+      "step": 2394
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.447975517890772e-05,
+      "loss": 0.3479,
+      "step": 2395
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4477401129943506e-05,
+      "loss": 0.329,
+      "step": 2396
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.447504708097929e-05,
+      "loss": 0.3437,
+      "step": 2397
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4472693032015067e-05,
+      "loss": 0.5861,
+      "step": 2398
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.447033898305085e-05,
+      "loss": 0.4432,
+      "step": 2399
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4467984934086634e-05,
+      "loss": 0.4509,
+      "step": 2400
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.446563088512241e-05,
+      "loss": 0.3798,
+      "step": 2401
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4463276836158196e-05,
+      "loss": 0.3453,
+      "step": 2402
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.446092278719397e-05,
+      "loss": 0.5449,
+      "step": 2403
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4458568738229757e-05,
+      "loss": 0.4543,
+      "step": 2404
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.445621468926554e-05,
+      "loss": 0.4095,
+      "step": 2405
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.445386064030132e-05,
+      "loss": 0.535,
+      "step": 2406
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.44515065913371e-05,
+      "loss": 0.3873,
+      "step": 2407
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4449152542372886e-05,
+      "loss": 0.246,
+      "step": 2408
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.444679849340866e-05,
+      "loss": 0.5748,
+      "step": 2409
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4444444444444447e-05,
+      "loss": 0.3782,
+      "step": 2410
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.444209039548023e-05,
+      "loss": 0.2427,
+      "step": 2411
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.443973634651601e-05,
+      "loss": 0.4131,
+      "step": 2412
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.443738229755179e-05,
+      "loss": 0.4256,
+      "step": 2413
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4435028248587575e-05,
+      "loss": 0.4035,
+      "step": 2414
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.443267419962335e-05,
+      "loss": 0.3472,
+      "step": 2415
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4430320150659137e-05,
+      "loss": 0.3293,
+      "step": 2416
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.442796610169492e-05,
+      "loss": 0.4341,
+      "step": 2417
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.44256120527307e-05,
+      "loss": 0.3846,
+      "step": 2418
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.442325800376648e-05,
+      "loss": 0.2845,
+      "step": 2419
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.442090395480226e-05,
+      "loss": 0.423,
+      "step": 2420
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.441854990583804e-05,
+      "loss": 0.311,
+      "step": 2421
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4416195856873827e-05,
+      "loss": 0.4163,
+      "step": 2422
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4413841807909604e-05,
+      "loss": 0.4626,
+      "step": 2423
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.441148775894539e-05,
+      "loss": 0.3255,
+      "step": 2424
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.440913370998117e-05,
+      "loss": 0.2891,
+      "step": 2425
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.440677966101695e-05,
+      "loss": 0.3549,
+      "step": 2426
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.440442561205273e-05,
+      "loss": 0.284,
+      "step": 2427
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4402071563088517e-05,
+      "loss": 0.4299,
+      "step": 2428
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4399717514124294e-05,
+      "loss": 0.3897,
+      "step": 2429
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.439736346516008e-05,
+      "loss": 0.4119,
+      "step": 2430
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.439500941619586e-05,
+      "loss": 0.4698,
+      "step": 2431
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.439265536723164e-05,
+      "loss": 0.3204,
+      "step": 2432
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.439030131826742e-05,
+      "loss": 0.4324,
+      "step": 2433
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4387947269303207e-05,
+      "loss": 0.5178,
+      "step": 2434
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4385593220338984e-05,
+      "loss": 0.4161,
+      "step": 2435
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.438323917137477e-05,
+      "loss": 0.433,
+      "step": 2436
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.438088512241055e-05,
+      "loss": 0.2371,
+      "step": 2437
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.437853107344633e-05,
+      "loss": 0.4247,
+      "step": 2438
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.437617702448211e-05,
+      "loss": 0.534,
+      "step": 2439
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.437382297551789e-05,
+      "loss": 0.4061,
+      "step": 2440
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4371468926553674e-05,
+      "loss": 0.3188,
+      "step": 2441
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.436911487758946e-05,
+      "loss": 0.4397,
+      "step": 2442
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4366760828625235e-05,
+      "loss": 0.5101,
+      "step": 2443
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.436440677966102e-05,
+      "loss": 0.3132,
+      "step": 2444
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.43620527306968e-05,
+      "loss": 0.2986,
+      "step": 2445
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.435969868173258e-05,
+      "loss": 0.421,
+      "step": 2446
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4357344632768364e-05,
+      "loss": 0.4035,
+      "step": 2447
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.435499058380415e-05,
+      "loss": 0.4246,
+      "step": 2448
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4352636534839925e-05,
+      "loss": 0.301,
+      "step": 2449
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.435028248587571e-05,
+      "loss": 0.4043,
+      "step": 2450
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.434792843691149e-05,
+      "loss": 0.4709,
+      "step": 2451
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.434557438794727e-05,
+      "loss": 0.3639,
+      "step": 2452
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4343220338983054e-05,
+      "loss": 0.4448,
+      "step": 2453
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.434086629001884e-05,
+      "loss": 0.3158,
+      "step": 2454
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4338512241054615e-05,
+      "loss": 0.5286,
+      "step": 2455
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.43361581920904e-05,
+      "loss": 0.406,
+      "step": 2456
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4333804143126176e-05,
+      "loss": 0.5293,
+      "step": 2457
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.433145009416196e-05,
+      "loss": 0.4521,
+      "step": 2458
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4329096045197744e-05,
+      "loss": 0.4247,
+      "step": 2459
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.432674199623352e-05,
+      "loss": 0.3678,
+      "step": 2460
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4324387947269305e-05,
+      "loss": 0.4607,
+      "step": 2461
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.432203389830509e-05,
+      "loss": 0.3729,
+      "step": 2462
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4319679849340866e-05,
+      "loss": 0.368,
+      "step": 2463
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.431732580037665e-05,
+      "loss": 0.3747,
+      "step": 2464
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4314971751412434e-05,
+      "loss": 0.4019,
+      "step": 2465
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.431261770244821e-05,
+      "loss": 0.3681,
+      "step": 2466
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4310263653483995e-05,
+      "loss": 0.4473,
+      "step": 2467
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.430790960451978e-05,
+      "loss": 0.3396,
+      "step": 2468
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4305555555555556e-05,
+      "loss": 0.348,
+      "step": 2469
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.430320150659134e-05,
+      "loss": 0.4891,
+      "step": 2470
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4300847457627124e-05,
+      "loss": 0.3164,
+      "step": 2471
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.42984934086629e-05,
+      "loss": 0.3628,
+      "step": 2472
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4296139359698685e-05,
+      "loss": 0.324,
+      "step": 2473
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.429378531073446e-05,
+      "loss": 0.3313,
+      "step": 2474
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4291431261770246e-05,
+      "loss": 0.3113,
+      "step": 2475
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.428907721280603e-05,
+      "loss": 0.3337,
+      "step": 2476
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.428672316384181e-05,
+      "loss": 0.4394,
+      "step": 2477
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.428436911487759e-05,
+      "loss": 0.516,
+      "step": 2478
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4282015065913375e-05,
+      "loss": 0.5217,
+      "step": 2479
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.427966101694915e-05,
+      "loss": 0.3838,
+      "step": 2480
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4277306967984936e-05,
+      "loss": 0.3996,
+      "step": 2481
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.427495291902072e-05,
+      "loss": 0.2792,
+      "step": 2482
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.42725988700565e-05,
+      "loss": 0.4346,
+      "step": 2483
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.427024482109228e-05,
+      "loss": 0.5248,
+      "step": 2484
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4267890772128065e-05,
+      "loss": 0.3685,
+      "step": 2485
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.426553672316384e-05,
+      "loss": 0.4237,
+      "step": 2486
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4263182674199626e-05,
+      "loss": 0.3983,
+      "step": 2487
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.426082862523541e-05,
+      "loss": 0.4248,
+      "step": 2488
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.425847457627119e-05,
+      "loss": 0.3387,
+      "step": 2489
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.425612052730697e-05,
+      "loss": 0.4235,
+      "step": 2490
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.425376647834275e-05,
+      "loss": 0.3432,
+      "step": 2491
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.425141242937854e-05,
+      "loss": 0.3675,
+      "step": 2492
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4249058380414316e-05,
+      "loss": 0.2681,
+      "step": 2493
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.424670433145009e-05,
+      "loss": 0.4392,
+      "step": 2494
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.424435028248588e-05,
+      "loss": 0.5015,
+      "step": 2495
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.424199623352166e-05,
+      "loss": 0.3107,
+      "step": 2496
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.423964218455744e-05,
+      "loss": 0.299,
+      "step": 2497
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.423728813559322e-05,
+      "loss": 0.4478,
+      "step": 2498
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4234934086629006e-05,
+      "loss": 0.3652,
+      "step": 2499
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.423258003766478e-05,
+      "loss": 0.4581,
+      "step": 2500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.423022598870057e-05,
+      "loss": 0.4598,
+      "step": 2501
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.422787193973635e-05,
+      "loss": 0.4576,
+      "step": 2502
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.422551789077213e-05,
+      "loss": 0.5318,
+      "step": 2503
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.422316384180791e-05,
+      "loss": 0.3469,
+      "step": 2504
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4220809792843696e-05,
+      "loss": 0.2829,
+      "step": 2505
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.421845574387947e-05,
+      "loss": 0.4041,
+      "step": 2506
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.421610169491526e-05,
+      "loss": 0.3993,
+      "step": 2507
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4213747645951034e-05,
+      "loss": 0.4652,
+      "step": 2508
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4211393596986825e-05,
+      "loss": 0.4373,
+      "step": 2509
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.42090395480226e-05,
+      "loss": 0.3948,
+      "step": 2510
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.420668549905838e-05,
+      "loss": 0.3879,
+      "step": 2511
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.420433145009417e-05,
+      "loss": 0.412,
+      "step": 2512
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.420197740112995e-05,
+      "loss": 0.2715,
+      "step": 2513
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4199623352165724e-05,
+      "loss": 0.3326,
+      "step": 2514
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.419726930320151e-05,
+      "loss": 0.3444,
+      "step": 2515
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.419491525423729e-05,
+      "loss": 0.4734,
+      "step": 2516
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.419256120527307e-05,
+      "loss": 0.3192,
+      "step": 2517
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.419020715630885e-05,
+      "loss": 0.2933,
+      "step": 2518
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.418785310734464e-05,
+      "loss": 0.3131,
+      "step": 2519
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4185499058380414e-05,
+      "loss": 0.4216,
+      "step": 2520
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.41831450094162e-05,
+      "loss": 0.4102,
+      "step": 2521
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.418079096045198e-05,
+      "loss": 0.4322,
+      "step": 2522
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.417843691148776e-05,
+      "loss": 0.4647,
+      "step": 2523
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.417608286252354e-05,
+      "loss": 0.4186,
+      "step": 2524
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.417372881355932e-05,
+      "loss": 0.5414,
+      "step": 2525
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.417137476459511e-05,
+      "loss": 0.3412,
+      "step": 2526
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.416902071563089e-05,
+      "loss": 0.6269,
+      "step": 2527
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4166666666666665e-05,
+      "loss": 0.3888,
+      "step": 2528
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4164312617702456e-05,
+      "loss": 0.3221,
+      "step": 2529
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.416195856873823e-05,
+      "loss": 0.337,
+      "step": 2530
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.415960451977401e-05,
+      "loss": 0.4342,
+      "step": 2531
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4157250470809794e-05,
+      "loss": 0.3287,
+      "step": 2532
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.415489642184558e-05,
+      "loss": 0.268,
+      "step": 2533
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4152542372881355e-05,
+      "loss": 0.4698,
+      "step": 2534
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.415018832391714e-05,
+      "loss": 0.3656,
+      "step": 2535
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.414783427495292e-05,
+      "loss": 0.4759,
+      "step": 2536
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.41454802259887e-05,
+      "loss": 0.4369,
+      "step": 2537
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4143126177024484e-05,
+      "loss": 0.3339,
+      "step": 2538
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.414077212806027e-05,
+      "loss": 0.3001,
+      "step": 2539
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.413841807909605e-05,
+      "loss": 0.4628,
+      "step": 2540
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.413606403013183e-05,
+      "loss": 0.5554,
+      "step": 2541
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4133709981167606e-05,
+      "loss": 0.3327,
+      "step": 2542
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.41313559322034e-05,
+      "loss": 0.4117,
+      "step": 2543
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4129001883239174e-05,
+      "loss": 0.3088,
+      "step": 2544
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.412664783427495e-05,
+      "loss": 0.4405,
+      "step": 2545
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.412429378531074e-05,
+      "loss": 0.429,
+      "step": 2546
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.412193973634652e-05,
+      "loss": 0.3285,
+      "step": 2547
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4119585687382296e-05,
+      "loss": 0.4089,
+      "step": 2548
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.411723163841808e-05,
+      "loss": 0.3651,
+      "step": 2549
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4114877589453864e-05,
+      "loss": 0.4012,
+      "step": 2550
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.411252354048964e-05,
+      "loss": 0.4519,
+      "step": 2551
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4110169491525425e-05,
+      "loss": 0.3381,
+      "step": 2552
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.410781544256121e-05,
+      "loss": 0.3207,
+      "step": 2553
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4105461393596986e-05,
+      "loss": 0.4094,
+      "step": 2554
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.410310734463277e-05,
+      "loss": 0.2846,
+      "step": 2555
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4100753295668554e-05,
+      "loss": 0.2985,
+      "step": 2556
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.409839924670434e-05,
+      "loss": 0.2752,
+      "step": 2557
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4096045197740115e-05,
+      "loss": 0.3632,
+      "step": 2558
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.409369114877589e-05,
+      "loss": 0.38,
+      "step": 2559
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.409133709981168e-05,
+      "loss": 0.3435,
+      "step": 2560
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.408898305084746e-05,
+      "loss": 0.3932,
+      "step": 2561
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.408662900188324e-05,
+      "loss": 0.5075,
+      "step": 2562
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.408427495291903e-05,
+      "loss": 0.3473,
+      "step": 2563
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4081920903954805e-05,
+      "loss": 0.4203,
+      "step": 2564
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.407956685499058e-05,
+      "loss": 0.3346,
+      "step": 2565
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4077212806026366e-05,
+      "loss": 0.3819,
+      "step": 2566
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.407485875706215e-05,
+      "loss": 0.4239,
+      "step": 2567
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.407250470809793e-05,
+      "loss": 0.3473,
+      "step": 2568
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.407015065913371e-05,
+      "loss": 0.3831,
+      "step": 2569
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4067796610169495e-05,
+      "loss": 0.4311,
+      "step": 2570
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.406544256120528e-05,
+      "loss": 0.4454,
+      "step": 2571
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4063088512241056e-05,
+      "loss": 0.3769,
+      "step": 2572
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.406073446327684e-05,
+      "loss": 0.4352,
+      "step": 2573
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4058380414312624e-05,
+      "loss": 0.307,
+      "step": 2574
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.40560263653484e-05,
+      "loss": 0.3328,
+      "step": 2575
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.405367231638418e-05,
+      "loss": 0.384,
+      "step": 2576
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.405131826741997e-05,
+      "loss": 0.3771,
+      "step": 2577
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4048964218455746e-05,
+      "loss": 0.3955,
+      "step": 2578
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.404661016949152e-05,
+      "loss": 0.4304,
+      "step": 2579
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4044256120527314e-05,
+      "loss": 0.3568,
+      "step": 2580
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.404190207156309e-05,
+      "loss": 0.3081,
+      "step": 2581
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.403954802259887e-05,
+      "loss": 0.3816,
+      "step": 2582
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.403719397363465e-05,
+      "loss": 0.4728,
+      "step": 2583
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4034839924670436e-05,
+      "loss": 0.3971,
+      "step": 2584
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.403248587570621e-05,
+      "loss": 0.3173,
+      "step": 2585
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4030131826742e-05,
+      "loss": 0.3217,
+      "step": 2586
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.402777777777778e-05,
+      "loss": 0.5156,
+      "step": 2587
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4025423728813565e-05,
+      "loss": 0.4125,
+      "step": 2588
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.402306967984934e-05,
+      "loss": 0.3521,
+      "step": 2589
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4020715630885126e-05,
+      "loss": 0.3301,
+      "step": 2590
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.401836158192091e-05,
+      "loss": 0.3294,
+      "step": 2591
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.401600753295669e-05,
+      "loss": 0.4171,
+      "step": 2592
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4013653483992464e-05,
+      "loss": 0.4895,
+      "step": 2593
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4011299435028255e-05,
+      "loss": 0.4266,
+      "step": 2594
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.400894538606403e-05,
+      "loss": 0.4507,
+      "step": 2595
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.400659133709981e-05,
+      "loss": 0.4167,
+      "step": 2596
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.40042372881356e-05,
+      "loss": 0.4652,
+      "step": 2597
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.400188323917138e-05,
+      "loss": 0.5024,
+      "step": 2598
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.3999529190207154e-05,
+      "loss": 0.4148,
+      "step": 2599
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.399717514124294e-05,
+      "loss": 0.2943,
+      "step": 2600
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.399482109227872e-05,
+      "loss": 0.3585,
+      "step": 2601
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.39924670433145e-05,
+      "loss": 0.457,
+      "step": 2602
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.399011299435028e-05,
+      "loss": 0.4661,
+      "step": 2603
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.398775894538607e-05,
+      "loss": 0.3609,
+      "step": 2604
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.398540489642185e-05,
+      "loss": 0.219,
+      "step": 2605
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.398305084745763e-05,
+      "loss": 0.3835,
+      "step": 2606
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.398069679849341e-05,
+      "loss": 0.3086,
+      "step": 2607
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.3978342749529196e-05,
+      "loss": 0.3565,
+      "step": 2608
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.397598870056497e-05,
+      "loss": 0.4523,
+      "step": 2609
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.397363465160075e-05,
+      "loss": 0.3384,
+      "step": 2610
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.397128060263654e-05,
+      "loss": 0.501,
+      "step": 2611
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.396892655367232e-05,
+      "loss": 0.3844,
+      "step": 2612
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3966572504708095e-05,
+      "loss": 0.389,
+      "step": 2613
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3964218455743886e-05,
+      "loss": 0.3833,
+      "step": 2614
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.396186440677966e-05,
+      "loss": 0.4833,
+      "step": 2615
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.395951035781544e-05,
+      "loss": 0.4174,
+      "step": 2616
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3957156308851224e-05,
+      "loss": 0.4949,
+      "step": 2617
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.395480225988701e-05,
+      "loss": 0.2707,
+      "step": 2618
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.395244821092279e-05,
+      "loss": 0.3634,
+      "step": 2619
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.395009416195857e-05,
+      "loss": 0.4254,
+      "step": 2620
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.394774011299435e-05,
+      "loss": 0.4675,
+      "step": 2621
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.394538606403014e-05,
+      "loss": 0.2153,
+      "step": 2622
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3943032015065914e-05,
+      "loss": 0.3898,
+      "step": 2623
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.39406779661017e-05,
+      "loss": 0.457,
+      "step": 2624
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.393832391713748e-05,
+      "loss": 0.2907,
+      "step": 2625
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.393596986817326e-05,
+      "loss": 0.3925,
+      "step": 2626
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3933615819209036e-05,
+      "loss": 0.3768,
+      "step": 2627
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.393126177024483e-05,
+      "loss": 0.4449,
+      "step": 2628
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3928907721280604e-05,
+      "loss": 0.4715,
+      "step": 2629
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.392655367231638e-05,
+      "loss": 0.4454,
+      "step": 2630
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.392419962335217e-05,
+      "loss": 0.3686,
+      "step": 2631
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.392184557438795e-05,
+      "loss": 0.4875,
+      "step": 2632
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3919491525423726e-05,
+      "loss": 0.4128,
+      "step": 2633
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.391713747645951e-05,
+      "loss": 0.4199,
+      "step": 2634
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3914783427495294e-05,
+      "loss": 0.3641,
+      "step": 2635
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.391242937853108e-05,
+      "loss": 0.3634,
+      "step": 2636
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3910075329566855e-05,
+      "loss": 0.4509,
+      "step": 2637
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.390772128060264e-05,
+      "loss": 0.4421,
+      "step": 2638
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.390536723163842e-05,
+      "loss": 0.3638,
+      "step": 2639
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.39030131826742e-05,
+      "loss": 0.4259,
+      "step": 2640
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3900659133709984e-05,
+      "loss": 0.4361,
+      "step": 2641
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.389830508474577e-05,
+      "loss": 0.4168,
+      "step": 2642
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3895951035781545e-05,
+      "loss": 0.3654,
+      "step": 2643
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.389359698681732e-05,
+      "loss": 0.4685,
+      "step": 2644
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.389124293785311e-05,
+      "loss": 0.3676,
+      "step": 2645
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.388888888888889e-05,
+      "loss": 0.3963,
+      "step": 2646
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.388653483992467e-05,
+      "loss": 0.3835,
+      "step": 2647
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.388418079096046e-05,
+      "loss": 0.3287,
+      "step": 2648
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3881826741996235e-05,
+      "loss": 0.2891,
+      "step": 2649
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.387947269303202e-05,
+      "loss": 0.2727,
+      "step": 2650
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3877118644067796e-05,
+      "loss": 0.3165,
+      "step": 2651
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.387476459510358e-05,
+      "loss": 0.2671,
+      "step": 2652
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3872410546139364e-05,
+      "loss": 0.3213,
+      "step": 2653
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.387005649717514e-05,
+      "loss": 0.3573,
+      "step": 2654
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3867702448210925e-05,
+      "loss": 0.3546,
+      "step": 2655
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.386534839924671e-05,
+      "loss": 0.4806,
+      "step": 2656
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3862994350282486e-05,
+      "loss": 0.3997,
+      "step": 2657
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.386064030131827e-05,
+      "loss": 0.235,
+      "step": 2658
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3858286252354054e-05,
+      "loss": 0.4329,
+      "step": 2659
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.385593220338983e-05,
+      "loss": 0.4057,
+      "step": 2660
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.385357815442561e-05,
+      "loss": 0.2987,
+      "step": 2661
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.38512241054614e-05,
+      "loss": 0.3087,
+      "step": 2662
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3848870056497176e-05,
+      "loss": 0.3073,
+      "step": 2663
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3846516007532954e-05,
+      "loss": 0.4368,
+      "step": 2664
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3844161958568744e-05,
+      "loss": 0.4395,
+      "step": 2665
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.384180790960452e-05,
+      "loss": 0.288,
+      "step": 2666
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3839453860640305e-05,
+      "loss": 0.3921,
+      "step": 2667
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.383709981167608e-05,
+      "loss": 0.4078,
+      "step": 2668
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3834745762711866e-05,
+      "loss": 0.2928,
+      "step": 2669
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.383239171374765e-05,
+      "loss": 0.3847,
+      "step": 2670
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.383003766478343e-05,
+      "loss": 0.2643,
+      "step": 2671
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.382768361581921e-05,
+      "loss": 0.5336,
+      "step": 2672
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3825329566854995e-05,
+      "loss": 0.3854,
+      "step": 2673
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.382297551789077e-05,
+      "loss": 0.4182,
+      "step": 2674
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3820621468926556e-05,
+      "loss": 0.4575,
+      "step": 2675
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.381826741996234e-05,
+      "loss": 0.2906,
+      "step": 2676
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.381591337099812e-05,
+      "loss": 0.1891,
+      "step": 2677
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.38135593220339e-05,
+      "loss": 0.2454,
+      "step": 2678
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3811205273069685e-05,
+      "loss": 0.3394,
+      "step": 2679
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.380885122410546e-05,
+      "loss": 0.4188,
+      "step": 2680
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.380649717514124e-05,
+      "loss": 0.4615,
+      "step": 2681
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.380414312617703e-05,
+      "loss": 0.3675,
+      "step": 2682
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.380178907721281e-05,
+      "loss": 0.3691,
+      "step": 2683
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.379943502824859e-05,
+      "loss": 0.3709,
+      "step": 2684
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.379708097928437e-05,
+      "loss": 0.6007,
+      "step": 2685
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.379472693032015e-05,
+      "loss": 0.4134,
+      "step": 2686
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3792372881355936e-05,
+      "loss": 0.361,
+      "step": 2687
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3790018832391713e-05,
+      "loss": 0.4143,
+      "step": 2688
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.37876647834275e-05,
+      "loss": 0.2538,
+      "step": 2689
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.378531073446328e-05,
+      "loss": 0.3965,
+      "step": 2690
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.378295668549906e-05,
+      "loss": 0.4714,
+      "step": 2691
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.378060263653484e-05,
+      "loss": 0.3849,
+      "step": 2692
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3778248587570626e-05,
+      "loss": 0.4317,
+      "step": 2693
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3775894538606403e-05,
+      "loss": 0.232,
+      "step": 2694
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.377354048964219e-05,
+      "loss": 0.3833,
+      "step": 2695
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.377118644067797e-05,
+      "loss": 0.3842,
+      "step": 2696
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.376883239171375e-05,
+      "loss": 0.3634,
+      "step": 2697
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.376647834274953e-05,
+      "loss": 0.362,
+      "step": 2698
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3764124293785316e-05,
+      "loss": 0.326,
+      "step": 2699
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3761770244821093e-05,
+      "loss": 0.4608,
+      "step": 2700
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.375941619585688e-05,
+      "loss": 0.4009,
+      "step": 2701
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3757062146892655e-05,
+      "loss": 0.448,
+      "step": 2702
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.375470809792844e-05,
+      "loss": 0.3205,
+      "step": 2703
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.375235404896422e-05,
+      "loss": 0.437,
+      "step": 2704
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.375e-05,
+      "loss": 0.4203,
+      "step": 2705
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3747645951035783e-05,
+      "loss": 0.4791,
+      "step": 2706
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.374529190207157e-05,
+      "loss": 0.4506,
+      "step": 2707
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3742937853107345e-05,
+      "loss": 0.269,
+      "step": 2708
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.374058380414313e-05,
+      "loss": 0.3523,
+      "step": 2709
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.373822975517891e-05,
+      "loss": 0.3804,
+      "step": 2710
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.373587570621469e-05,
+      "loss": 0.4727,
+      "step": 2711
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3733521657250473e-05,
+      "loss": 0.3734,
+      "step": 2712
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.373116760828626e-05,
+      "loss": 0.6498,
+      "step": 2713
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3728813559322035e-05,
+      "loss": 0.4472,
+      "step": 2714
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.372645951035782e-05,
+      "loss": 0.6315,
+      "step": 2715
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.37241054613936e-05,
+      "loss": 0.3465,
+      "step": 2716
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.372175141242938e-05,
+      "loss": 0.5125,
+      "step": 2717
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3719397363465163e-05,
+      "loss": 0.4905,
+      "step": 2718
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.371704331450094e-05,
+      "loss": 0.3502,
+      "step": 2719
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3714689265536725e-05,
+      "loss": 0.5483,
+      "step": 2720
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.371233521657251e-05,
+      "loss": 0.3425,
+      "step": 2721
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3709981167608286e-05,
+      "loss": 0.3669,
+      "step": 2722
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.370762711864407e-05,
+      "loss": 0.3955,
+      "step": 2723
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3705273069679853e-05,
+      "loss": 0.3503,
+      "step": 2724
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.370291902071563e-05,
+      "loss": 0.3342,
+      "step": 2725
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3700564971751415e-05,
+      "loss": 0.3505,
+      "step": 2726
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.36982109227872e-05,
+      "loss": 0.3601,
+      "step": 2727
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3695856873822976e-05,
+      "loss": 0.347,
+      "step": 2728
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.369350282485876e-05,
+      "loss": 0.3941,
+      "step": 2729
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3691148775894543e-05,
+      "loss": 0.423,
+      "step": 2730
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.368879472693032e-05,
+      "loss": 0.3504,
+      "step": 2731
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3686440677966105e-05,
+      "loss": 0.2657,
+      "step": 2732
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.368408662900189e-05,
+      "loss": 0.3023,
+      "step": 2733
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3681732580037666e-05,
+      "loss": 0.4181,
+      "step": 2734
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.367937853107345e-05,
+      "loss": 0.4667,
+      "step": 2735
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.367702448210923e-05,
+      "loss": 0.4623,
+      "step": 2736
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.367467043314501e-05,
+      "loss": 0.339,
+      "step": 2737
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3672316384180795e-05,
+      "loss": 0.4737,
+      "step": 2738
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.366996233521657e-05,
+      "loss": 0.3666,
+      "step": 2739
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3667608286252356e-05,
+      "loss": 0.4441,
+      "step": 2740
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.366525423728814e-05,
+      "loss": 0.2703,
+      "step": 2741
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.366290018832392e-05,
+      "loss": 0.4418,
+      "step": 2742
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.36605461393597e-05,
+      "loss": 0.37,
+      "step": 2743
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3658192090395485e-05,
+      "loss": 0.3745,
+      "step": 2744
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.365583804143126e-05,
+      "loss": 0.4703,
+      "step": 2745
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3653483992467046e-05,
+      "loss": 0.3782,
+      "step": 2746
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.365112994350283e-05,
+      "loss": 0.2435,
+      "step": 2747
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.364877589453861e-05,
+      "loss": 0.3581,
+      "step": 2748
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.364642184557439e-05,
+      "loss": 0.4568,
+      "step": 2749
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3644067796610175e-05,
+      "loss": 0.4073,
+      "step": 2750
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.364171374764595e-05,
+      "loss": 0.3944,
+      "step": 2751
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3639359698681736e-05,
+      "loss": 0.3408,
+      "step": 2752
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.363700564971752e-05,
+      "loss": 0.4593,
+      "step": 2753
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.36346516007533e-05,
+      "loss": 0.3584,
+      "step": 2754
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.363229755178908e-05,
+      "loss": 0.4258,
+      "step": 2755
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.362994350282486e-05,
+      "loss": 0.3994,
+      "step": 2756
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.362758945386064e-05,
+      "loss": 0.4523,
+      "step": 2757
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3625235404896426e-05,
+      "loss": 0.4487,
+      "step": 2758
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.36228813559322e-05,
+      "loss": 0.4654,
+      "step": 2759
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.362052730696799e-05,
+      "loss": 0.2992,
+      "step": 2760
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.361817325800377e-05,
+      "loss": 0.2245,
+      "step": 2761
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.361581920903955e-05,
+      "loss": 0.3528,
+      "step": 2762
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.361346516007533e-05,
+      "loss": 0.3912,
+      "step": 2763
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3611111111111116e-05,
+      "loss": 0.3978,
+      "step": 2764
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.360875706214689e-05,
+      "loss": 0.3248,
+      "step": 2765
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.360640301318268e-05,
+      "loss": 0.3101,
+      "step": 2766
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.360404896421846e-05,
+      "loss": 0.3338,
+      "step": 2767
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.360169491525424e-05,
+      "loss": 0.3384,
+      "step": 2768
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.359934086629002e-05,
+      "loss": 0.4611,
+      "step": 2769
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3596986817325806e-05,
+      "loss": 0.3814,
+      "step": 2770
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.359463276836158e-05,
+      "loss": 0.4229,
+      "step": 2771
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.359227871939737e-05,
+      "loss": 0.3707,
+      "step": 2772
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3589924670433144e-05,
+      "loss": 0.4887,
+      "step": 2773
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.358757062146893e-05,
+      "loss": 0.437,
+      "step": 2774
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.358521657250471e-05,
+      "loss": 0.4191,
+      "step": 2775
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.358286252354049e-05,
+      "loss": 0.4409,
+      "step": 2776
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.358050847457627e-05,
+      "loss": 0.5313,
+      "step": 2777
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.357815442561206e-05,
+      "loss": 0.3413,
+      "step": 2778
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3575800376647834e-05,
+      "loss": 0.364,
+      "step": 2779
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.357344632768362e-05,
+      "loss": 0.2957,
+      "step": 2780
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.35710922787194e-05,
+      "loss": 0.4519,
+      "step": 2781
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.356873822975518e-05,
+      "loss": 0.33,
+      "step": 2782
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.356638418079096e-05,
+      "loss": 0.4713,
+      "step": 2783
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.356403013182675e-05,
+      "loss": 0.3882,
+      "step": 2784
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3561676082862524e-05,
+      "loss": 0.3619,
+      "step": 2785
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.355932203389831e-05,
+      "loss": 0.33,
+      "step": 2786
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.355696798493409e-05,
+      "loss": 0.351,
+      "step": 2787
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.355461393596987e-05,
+      "loss": 0.4753,
+      "step": 2788
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.355225988700565e-05,
+      "loss": 0.3451,
+      "step": 2789
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.354990583804143e-05,
+      "loss": 0.2758,
+      "step": 2790
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3547551789077214e-05,
+      "loss": 0.4755,
+      "step": 2791
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3545197740113e-05,
+      "loss": 0.3446,
+      "step": 2792
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3542843691148775e-05,
+      "loss": 0.3439,
+      "step": 2793
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.354048964218456e-05,
+      "loss": 0.384,
+      "step": 2794
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.353813559322034e-05,
+      "loss": 0.4188,
+      "step": 2795
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.353578154425612e-05,
+      "loss": 0.3835,
+      "step": 2796
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3533427495291904e-05,
+      "loss": 0.3159,
+      "step": 2797
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.353107344632769e-05,
+      "loss": 0.4199,
+      "step": 2798
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3528719397363465e-05,
+      "loss": 0.3017,
+      "step": 2799
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.352636534839925e-05,
+      "loss": 0.443,
+      "step": 2800
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.352401129943503e-05,
+      "loss": 0.4078,
+      "step": 2801
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.352165725047081e-05,
+      "loss": 0.4473,
+      "step": 2802
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3519303201506594e-05,
+      "loss": 0.3763,
+      "step": 2803
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.351694915254238e-05,
+      "loss": 0.3308,
+      "step": 2804
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3514595103578155e-05,
+      "loss": 0.4383,
+      "step": 2805
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.351224105461394e-05,
+      "loss": 0.3675,
+      "step": 2806
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3509887005649716e-05,
+      "loss": 0.4266,
+      "step": 2807
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3507532956685507e-05,
+      "loss": 0.3707,
+      "step": 2808
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3505178907721284e-05,
+      "loss": 0.4003,
+      "step": 2809
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.350282485875706e-05,
+      "loss": 0.5227,
+      "step": 2810
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3500470809792845e-05,
+      "loss": 0.3981,
+      "step": 2811
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.349811676082863e-05,
+      "loss": 0.3579,
+      "step": 2812
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3495762711864406e-05,
+      "loss": 0.339,
+      "step": 2813
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.349340866290019e-05,
+      "loss": 0.3777,
+      "step": 2814
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3491054613935974e-05,
+      "loss": 0.3531,
+      "step": 2815
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.348870056497175e-05,
+      "loss": 0.4414,
+      "step": 2816
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3486346516007535e-05,
+      "loss": 0.3231,
+      "step": 2817
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.348399246704332e-05,
+      "loss": 0.3201,
+      "step": 2818
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3481638418079096e-05,
+      "loss": 0.2447,
+      "step": 2819
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.347928436911488e-05,
+      "loss": 0.3243,
+      "step": 2820
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3476930320150664e-05,
+      "loss": 0.3811,
+      "step": 2821
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.347457627118644e-05,
+      "loss": 0.3177,
+      "step": 2822
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3472222222222225e-05,
+      "loss": 0.4294,
+      "step": 2823
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3469868173258e-05,
+      "loss": 0.6724,
+      "step": 2824
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.346751412429379e-05,
+      "loss": 0.5909,
+      "step": 2825
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.346516007532957e-05,
+      "loss": 0.4208,
+      "step": 2826
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.346280602636535e-05,
+      "loss": 0.3509,
+      "step": 2827
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.346045197740114e-05,
+      "loss": 0.2323,
+      "step": 2828
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3458097928436915e-05,
+      "loss": 0.3195,
+      "step": 2829
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.345574387947269e-05,
+      "loss": 0.2976,
+      "step": 2830
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3453389830508476e-05,
+      "loss": 0.2969,
+      "step": 2831
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.345103578154426e-05,
+      "loss": 0.3182,
+      "step": 2832
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.344868173258004e-05,
+      "loss": 0.4547,
+      "step": 2833
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.344632768361582e-05,
+      "loss": 0.3712,
+      "step": 2834
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3443973634651605e-05,
+      "loss": 0.4508,
+      "step": 2835
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.344161958568738e-05,
+      "loss": 0.5325,
+      "step": 2836
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3439265536723166e-05,
+      "loss": 0.4184,
+      "step": 2837
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.343691148775895e-05,
+      "loss": 0.4019,
+      "step": 2838
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.343455743879473e-05,
+      "loss": 0.4675,
+      "step": 2839
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.343220338983051e-05,
+      "loss": 0.4616,
+      "step": 2840
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.342984934086629e-05,
+      "loss": 0.3186,
+      "step": 2841
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.342749529190208e-05,
+      "loss": 0.4555,
+      "step": 2842
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3425141242937856e-05,
+      "loss": 0.4151,
+      "step": 2843
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.342278719397363e-05,
+      "loss": 0.3747,
+      "step": 2844
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3420433145009424e-05,
+      "loss": 0.4874,
+      "step": 2845
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.34180790960452e-05,
+      "loss": 0.4061,
+      "step": 2846
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.341572504708098e-05,
+      "loss": 0.4052,
+      "step": 2847
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.341337099811676e-05,
+      "loss": 0.4152,
+      "step": 2848
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3411016949152546e-05,
+      "loss": 0.3949,
+      "step": 2849
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.340866290018832e-05,
+      "loss": 0.3969,
+      "step": 2850
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.340630885122411e-05,
+      "loss": 0.4013,
+      "step": 2851
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.340395480225989e-05,
+      "loss": 0.4207,
+      "step": 2852
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.340160075329567e-05,
+      "loss": 0.4492,
+      "step": 2853
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.339924670433145e-05,
+      "loss": 0.356,
+      "step": 2854
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3396892655367236e-05,
+      "loss": 0.3917,
+      "step": 2855
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.339453860640302e-05,
+      "loss": 0.2806,
+      "step": 2856
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.33921845574388e-05,
+      "loss": 0.4238,
+      "step": 2857
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3389830508474574e-05,
+      "loss": 0.2793,
+      "step": 2858
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3387476459510365e-05,
+      "loss": 0.3003,
+      "step": 2859
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.338512241054614e-05,
+      "loss": 0.3824,
+      "step": 2860
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.338276836158192e-05,
+      "loss": 0.416,
+      "step": 2861
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.338041431261771e-05,
+      "loss": 0.3982,
+      "step": 2862
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.337806026365349e-05,
+      "loss": 0.4288,
+      "step": 2863
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3375706214689264e-05,
+      "loss": 0.4661,
+      "step": 2864
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.337335216572505e-05,
+      "loss": 0.4377,
+      "step": 2865
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.337099811676083e-05,
+      "loss": 0.2414,
+      "step": 2866
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.336864406779661e-05,
+      "loss": 0.3788,
+      "step": 2867
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.336629001883239e-05,
+      "loss": 0.2985,
+      "step": 2868
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.336393596986818e-05,
+      "loss": 0.4266,
+      "step": 2869
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3361581920903954e-05,
+      "loss": 0.3323,
+      "step": 2870
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.335922787193974e-05,
+      "loss": 0.4119,
+      "step": 2871
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.335687382297552e-05,
+      "loss": 0.4305,
+      "step": 2872
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3354519774011306e-05,
+      "loss": 0.4047,
+      "step": 2873
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.335216572504708e-05,
+      "loss": 0.3039,
+      "step": 2874
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.334981167608286e-05,
+      "loss": 0.3664,
+      "step": 2875
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.334745762711865e-05,
+      "loss": 0.4417,
+      "step": 2876
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.334510357815443e-05,
+      "loss": 0.3368,
+      "step": 2877
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3342749529190205e-05,
+      "loss": 0.5073,
+      "step": 2878
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3340395480225996e-05,
+      "loss": 0.4984,
+      "step": 2879
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.333804143126177e-05,
+      "loss": 0.4926,
+      "step": 2880
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.333568738229755e-05,
+      "loss": 0.5094,
+      "step": 2881
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3333333333333334e-05,
+      "loss": 0.3616,
+      "step": 2882
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.333097928436912e-05,
+      "loss": 0.4341,
+      "step": 2883
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3328625235404895e-05,
+      "loss": 0.3329,
+      "step": 2884
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.332627118644068e-05,
+      "loss": 0.3458,
+      "step": 2885
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.332391713747646e-05,
+      "loss": 0.3199,
+      "step": 2886
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.332156308851225e-05,
+      "loss": 0.4483,
+      "step": 2887
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.3319209039548024e-05,
+      "loss": 0.5369,
+      "step": 2888
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.331685499058381e-05,
+      "loss": 0.3723,
+      "step": 2889
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.331450094161959e-05,
+      "loss": 0.4155,
+      "step": 2890
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.331214689265537e-05,
+      "loss": 0.4595,
+      "step": 2891
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.3309792843691146e-05,
+      "loss": 0.3393,
+      "step": 2892
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.330743879472694e-05,
+      "loss": 0.3387,
+      "step": 2893
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.3305084745762714e-05,
+      "loss": 0.3632,
+      "step": 2894
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.330273069679849e-05,
+      "loss": 0.3668,
+      "step": 2895
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.330037664783428e-05,
+      "loss": 0.3314,
+      "step": 2896
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.329802259887006e-05,
+      "loss": 0.4337,
+      "step": 2897
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.3295668549905836e-05,
+      "loss": 0.3536,
+      "step": 2898
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.329331450094162e-05,
+      "loss": 0.3671,
+      "step": 2899
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.3290960451977404e-05,
+      "loss": 0.4133,
+      "step": 2900
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.328860640301318e-05,
+      "loss": 0.3728,
+      "step": 2901
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.3286252354048965e-05,
+      "loss": 0.3031,
+      "step": 2902
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.328389830508475e-05,
+      "loss": 0.3042,
+      "step": 2903
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.328154425612053e-05,
+      "loss": 0.3295,
+      "step": 2904
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.327919020715631e-05,
+      "loss": 0.328,
+      "step": 2905
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.3276836158192094e-05,
+      "loss": 0.3713,
+      "step": 2906
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.327448210922788e-05,
+      "loss": 0.2174,
+      "step": 2907
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3272128060263655e-05,
+      "loss": 0.3165,
+      "step": 2908
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.326977401129943e-05,
+      "loss": 0.4014,
+      "step": 2909
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.326741996233522e-05,
+      "loss": 0.54,
+      "step": 2910
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3265065913371e-05,
+      "loss": 0.3928,
+      "step": 2911
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.326271186440678e-05,
+      "loss": 0.3109,
+      "step": 2912
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.326035781544257e-05,
+      "loss": 0.3118,
+      "step": 2913
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3258003766478345e-05,
+      "loss": 0.2205,
+      "step": 2914
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.325564971751412e-05,
+      "loss": 0.4318,
+      "step": 2915
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3253295668549906e-05,
+      "loss": 0.2334,
+      "step": 2916
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.325094161958569e-05,
+      "loss": 0.5417,
+      "step": 2917
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.324858757062147e-05,
+      "loss": 0.3744,
+      "step": 2918
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.324623352165725e-05,
+      "loss": 0.5058,
+      "step": 2919
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3243879472693035e-05,
+      "loss": 0.4753,
+      "step": 2920
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.324152542372882e-05,
+      "loss": 0.5681,
+      "step": 2921
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3239171374764596e-05,
+      "loss": 0.4117,
+      "step": 2922
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.323681732580038e-05,
+      "loss": 0.3476,
+      "step": 2923
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3234463276836164e-05,
+      "loss": 0.3977,
+      "step": 2924
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.323210922787194e-05,
+      "loss": 0.2429,
+      "step": 2925
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.322975517890772e-05,
+      "loss": 0.3139,
+      "step": 2926
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.322740112994351e-05,
+      "loss": 0.4687,
+      "step": 2927
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3225047080979286e-05,
+      "loss": 0.2964,
+      "step": 2928
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.322269303201506e-05,
+      "loss": 0.399,
+      "step": 2929
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3220338983050854e-05,
+      "loss": 0.5041,
+      "step": 2930
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.321798493408663e-05,
+      "loss": 0.32,
+      "step": 2931
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.321563088512241e-05,
+      "loss": 0.3986,
+      "step": 2932
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.321327683615819e-05,
+      "loss": 0.3871,
+      "step": 2933
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3210922787193976e-05,
+      "loss": 0.3367,
+      "step": 2934
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.320856873822976e-05,
+      "loss": 0.4201,
+      "step": 2935
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.320621468926554e-05,
+      "loss": 0.3991,
+      "step": 2936
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.320386064030132e-05,
+      "loss": 0.4082,
+      "step": 2937
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3201506591337105e-05,
+      "loss": 0.4342,
+      "step": 2938
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.319915254237288e-05,
+      "loss": 0.4364,
+      "step": 2939
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3196798493408666e-05,
+      "loss": 0.3281,
+      "step": 2940
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.319444444444445e-05,
+      "loss": 0.361,
+      "step": 2941
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.319209039548023e-05,
+      "loss": 0.3721,
+      "step": 2942
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3189736346516004e-05,
+      "loss": 0.4278,
+      "step": 2943
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3187382297551795e-05,
+      "loss": 0.3693,
+      "step": 2944
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.318502824858757e-05,
+      "loss": 0.3412,
+      "step": 2945
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.318267419962335e-05,
+      "loss": 0.354,
+      "step": 2946
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.318032015065914e-05,
+      "loss": 0.3952,
+      "step": 2947
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.317796610169492e-05,
+      "loss": 0.5121,
+      "step": 2948
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3175612052730694e-05,
+      "loss": 0.3728,
+      "step": 2949
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.317325800376648e-05,
+      "loss": 0.393,
+      "step": 2950
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.317090395480226e-05,
+      "loss": 0.2488,
+      "step": 2951
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3168549905838046e-05,
+      "loss": 0.3888,
+      "step": 2952
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.316619585687382e-05,
+      "loss": 0.3708,
+      "step": 2953
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.316384180790961e-05,
+      "loss": 0.3309,
+      "step": 2954
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.316148775894539e-05,
+      "loss": 0.4505,
+      "step": 2955
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.315913370998117e-05,
+      "loss": 0.3483,
+      "step": 2956
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.315677966101695e-05,
+      "loss": 0.2707,
+      "step": 2957
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3154425612052736e-05,
+      "loss": 0.3345,
+      "step": 2958
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.315207156308851e-05,
+      "loss": 0.4333,
+      "step": 2959
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.314971751412429e-05,
+      "loss": 0.3956,
+      "step": 2960
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.314736346516008e-05,
+      "loss": 0.3941,
+      "step": 2961
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.314500941619586e-05,
+      "loss": 0.3502,
+      "step": 2962
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3142655367231635e-05,
+      "loss": 0.3262,
+      "step": 2963
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3140301318267426e-05,
+      "loss": 0.2927,
+      "step": 2964
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.31379472693032e-05,
+      "loss": 0.3721,
+      "step": 2965
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.313559322033899e-05,
+      "loss": 0.4059,
+      "step": 2966
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3133239171374764e-05,
+      "loss": 0.4469,
+      "step": 2967
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.313088512241055e-05,
+      "loss": 0.322,
+      "step": 2968
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.312853107344633e-05,
+      "loss": 0.4238,
+      "step": 2969
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.312617702448211e-05,
+      "loss": 0.3898,
+      "step": 2970
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.312382297551789e-05,
+      "loss": 0.3889,
+      "step": 2971
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.312146892655368e-05,
+      "loss": 0.3694,
+      "step": 2972
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3119114877589454e-05,
+      "loss": 0.5344,
+      "step": 2973
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.311676082862524e-05,
+      "loss": 0.3574,
+      "step": 2974
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.311440677966102e-05,
+      "loss": 0.3842,
+      "step": 2975
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.31120527306968e-05,
+      "loss": 0.3554,
+      "step": 2976
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3109698681732576e-05,
+      "loss": 0.4321,
+      "step": 2977
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.310734463276837e-05,
+      "loss": 0.3328,
+      "step": 2978
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3104990583804144e-05,
+      "loss": 0.2957,
+      "step": 2979
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.310263653483992e-05,
+      "loss": 0.4516,
+      "step": 2980
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.310028248587571e-05,
+      "loss": 0.3525,
+      "step": 2981
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.309792843691149e-05,
+      "loss": 0.4227,
+      "step": 2982
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.309557438794727e-05,
+      "loss": 0.3991,
+      "step": 2983
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.309322033898305e-05,
+      "loss": 0.5274,
+      "step": 2984
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3090866290018834e-05,
+      "loss": 0.406,
+      "step": 2985
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.308851224105462e-05,
+      "loss": 0.3351,
+      "step": 2986
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3086158192090395e-05,
+      "loss": 0.4066,
+      "step": 2987
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.308380414312618e-05,
+      "loss": 0.3484,
+      "step": 2988
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.308145009416196e-05,
+      "loss": 0.3788,
+      "step": 2989
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.307909604519774e-05,
+      "loss": 0.4059,
+      "step": 2990
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3076741996233524e-05,
+      "loss": 0.3833,
+      "step": 2991
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.307438794726931e-05,
+      "loss": 0.4194,
+      "step": 2992
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.3072033898305085e-05,
+      "loss": 0.4089,
+      "step": 2993
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.306967984934087e-05,
+      "loss": 0.3486,
+      "step": 2994
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.306732580037665e-05,
+      "loss": 0.5285,
+      "step": 2995
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.306497175141243e-05,
+      "loss": 0.465,
+      "step": 2996
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.306261770244821e-05,
+      "loss": 0.3208,
+      "step": 2997
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.3060263653484e-05,
+      "loss": 0.4546,
+      "step": 2998
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.3057909604519775e-05,
+      "loss": 0.3511,
+      "step": 2999
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.305555555555556e-05,
+      "loss": 0.3189,
+      "step": 3000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.3053201506591336e-05,
+      "loss": 0.3582,
+      "step": 3001
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.305084745762712e-05,
+      "loss": 0.4247,
+      "step": 3002
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.3048493408662904e-05,
+      "loss": 0.3524,
+      "step": 3003
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.304613935969868e-05,
+      "loss": 0.3298,
+      "step": 3004
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.3043785310734465e-05,
+      "loss": 0.4326,
+      "step": 3005
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.304143126177025e-05,
+      "loss": 0.5143,
+      "step": 3006
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.3039077212806026e-05,
+      "loss": 0.4599,
+      "step": 3007
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.303672316384181e-05,
+      "loss": 0.3859,
+      "step": 3008
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.3034369114877594e-05,
+      "loss": 0.3689,
+      "step": 3009
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.303201506591337e-05,
+      "loss": 0.386,
+      "step": 3010
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.3029661016949155e-05,
+      "loss": 0.3319,
+      "step": 3011
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.302730696798494e-05,
+      "loss": 0.4706,
+      "step": 3012
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.3024952919020716e-05,
+      "loss": 0.379,
+      "step": 3013
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.30225988700565e-05,
+      "loss": 0.3475,
+      "step": 3014
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.3020244821092284e-05,
+      "loss": 0.4698,
+      "step": 3015
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.301789077212806e-05,
+      "loss": 0.2794,
+      "step": 3016
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.3015536723163845e-05,
+      "loss": 0.2982,
+      "step": 3017
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.301318267419962e-05,
+      "loss": 0.4764,
+      "step": 3018
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.3010828625235406e-05,
+      "loss": 0.32,
+      "step": 3019
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.300847457627119e-05,
+      "loss": 0.4469,
+      "step": 3020
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.300612052730697e-05,
+      "loss": 0.3125,
+      "step": 3021
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.300376647834275e-05,
+      "loss": 0.2336,
+      "step": 3022
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.3001412429378535e-05,
+      "loss": 0.274,
+      "step": 3023
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.299905838041431e-05,
+      "loss": 0.331,
+      "step": 3024
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2996704331450096e-05,
+      "loss": 0.3001,
+      "step": 3025
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.299435028248588e-05,
+      "loss": 0.353,
+      "step": 3026
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.299199623352166e-05,
+      "loss": 0.4307,
+      "step": 3027
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.298964218455744e-05,
+      "loss": 0.3507,
+      "step": 3028
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2987288135593225e-05,
+      "loss": 0.3548,
+      "step": 3029
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2984934086629e-05,
+      "loss": 0.397,
+      "step": 3030
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2982580037664786e-05,
+      "loss": 0.3303,
+      "step": 3031
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.298022598870057e-05,
+      "loss": 0.3268,
+      "step": 3032
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.297787193973635e-05,
+      "loss": 0.5329,
+      "step": 3033
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.297551789077213e-05,
+      "loss": 0.3059,
+      "step": 3034
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.297316384180791e-05,
+      "loss": 0.3185,
+      "step": 3035
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.297080979284369e-05,
+      "loss": 0.4519,
+      "step": 3036
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2968455743879476e-05,
+      "loss": 0.271,
+      "step": 3037
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2966101694915254e-05,
+      "loss": 0.4813,
+      "step": 3038
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.296374764595104e-05,
+      "loss": 0.456,
+      "step": 3039
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.296139359698682e-05,
+      "loss": 0.3136,
+      "step": 3040
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.29590395480226e-05,
+      "loss": 0.3373,
+      "step": 3041
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.295668549905838e-05,
+      "loss": 0.3301,
+      "step": 3042
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2954331450094166e-05,
+      "loss": 0.3102,
+      "step": 3043
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2951977401129944e-05,
+      "loss": 0.3176,
+      "step": 3044
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.294962335216573e-05,
+      "loss": 0.3778,
+      "step": 3045
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.294726930320151e-05,
+      "loss": 0.3175,
+      "step": 3046
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.294491525423729e-05,
+      "loss": 0.2687,
+      "step": 3047
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.294256120527307e-05,
+      "loss": 0.3953,
+      "step": 3048
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2940207156308856e-05,
+      "loss": 0.2509,
+      "step": 3049
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2937853107344634e-05,
+      "loss": 0.482,
+      "step": 3050
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.293549905838042e-05,
+      "loss": 0.3088,
+      "step": 3051
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2933145009416195e-05,
+      "loss": 0.4305,
+      "step": 3052
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.293079096045198e-05,
+      "loss": 0.33,
+      "step": 3053
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.292843691148776e-05,
+      "loss": 0.5258,
+      "step": 3054
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.292608286252354e-05,
+      "loss": 0.3442,
+      "step": 3055
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2923728813559324e-05,
+      "loss": 0.3206,
+      "step": 3056
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.292137476459511e-05,
+      "loss": 0.2041,
+      "step": 3057
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2919020715630885e-05,
+      "loss": 0.2876,
+      "step": 3058
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.291666666666667e-05,
+      "loss": 0.3355,
+      "step": 3059
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.291431261770245e-05,
+      "loss": 0.3411,
+      "step": 3060
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.291195856873823e-05,
+      "loss": 0.3511,
+      "step": 3061
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2909604519774014e-05,
+      "loss": 0.3459,
+      "step": 3062
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.29072504708098e-05,
+      "loss": 0.4088,
+      "step": 3063
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2904896421845575e-05,
+      "loss": 0.3658,
+      "step": 3064
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.290254237288136e-05,
+      "loss": 0.404,
+      "step": 3065
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.290018832391714e-05,
+      "loss": 0.2589,
+      "step": 3066
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.289783427495292e-05,
+      "loss": 0.3315,
+      "step": 3067
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2895480225988704e-05,
+      "loss": 0.4433,
+      "step": 3068
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.289312617702449e-05,
+      "loss": 0.4018,
+      "step": 3069
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2890772128060265e-05,
+      "loss": 0.2432,
+      "step": 3070
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.288841807909605e-05,
+      "loss": 0.3892,
+      "step": 3071
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2886064030131826e-05,
+      "loss": 0.3804,
+      "step": 3072
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.288370998116761e-05,
+      "loss": 0.4063,
+      "step": 3073
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2881355932203394e-05,
+      "loss": 0.3925,
+      "step": 3074
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.287900188323917e-05,
+      "loss": 0.4649,
+      "step": 3075
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2876647834274955e-05,
+      "loss": 0.2829,
+      "step": 3076
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.287429378531074e-05,
+      "loss": 0.3508,
+      "step": 3077
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2871939736346516e-05,
+      "loss": 0.3055,
+      "step": 3078
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.28695856873823e-05,
+      "loss": 0.3725,
+      "step": 3079
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2867231638418084e-05,
+      "loss": 0.4033,
+      "step": 3080
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.286487758945386e-05,
+      "loss": 0.2621,
+      "step": 3081
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2862523540489645e-05,
+      "loss": 0.364,
+      "step": 3082
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.286016949152543e-05,
+      "loss": 0.3372,
+      "step": 3083
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2857815442561206e-05,
+      "loss": 0.3313,
+      "step": 3084
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.285546139359699e-05,
+      "loss": 0.2142,
+      "step": 3085
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2853107344632774e-05,
+      "loss": 0.313,
+      "step": 3086
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.285075329566855e-05,
+      "loss": 0.4062,
+      "step": 3087
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2848399246704335e-05,
+      "loss": 0.4272,
+      "step": 3088
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.284604519774011e-05,
+      "loss": 0.3644,
+      "step": 3089
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2843691148775896e-05,
+      "loss": 0.4755,
+      "step": 3090
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.284133709981168e-05,
+      "loss": 0.3009,
+      "step": 3091
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.283898305084746e-05,
+      "loss": 0.32,
+      "step": 3092
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.283662900188324e-05,
+      "loss": 0.3327,
+      "step": 3093
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2834274952919025e-05,
+      "loss": 0.3836,
+      "step": 3094
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.28319209039548e-05,
+      "loss": 0.5186,
+      "step": 3095
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2829566854990586e-05,
+      "loss": 0.4268,
+      "step": 3096
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.282721280602637e-05,
+      "loss": 0.3328,
+      "step": 3097
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.282485875706215e-05,
+      "loss": 0.3774,
+      "step": 3098
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.282250470809793e-05,
+      "loss": 0.4055,
+      "step": 3099
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2820150659133715e-05,
+      "loss": 0.3739,
+      "step": 3100
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.281779661016949e-05,
+      "loss": 0.3447,
+      "step": 3101
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2815442561205276e-05,
+      "loss": 0.2697,
+      "step": 3102
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.281308851224106e-05,
+      "loss": 0.2621,
+      "step": 3103
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.281073446327684e-05,
+      "loss": 0.3755,
+      "step": 3104
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.280838041431262e-05,
+      "loss": 0.3555,
+      "step": 3105
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.28060263653484e-05,
+      "loss": 0.3126,
+      "step": 3106
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.280367231638418e-05,
+      "loss": 0.3742,
+      "step": 3107
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2801318267419966e-05,
+      "loss": 0.3558,
+      "step": 3108
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.279896421845574e-05,
+      "loss": 0.4095,
+      "step": 3109
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.279661016949153e-05,
+      "loss": 0.3707,
+      "step": 3110
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.279425612052731e-05,
+      "loss": 0.3236,
+      "step": 3111
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.279190207156309e-05,
+      "loss": 0.3311,
+      "step": 3112
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.278954802259887e-05,
+      "loss": 0.3742,
+      "step": 3113
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2787193973634656e-05,
+      "loss": 0.3563,
+      "step": 3114
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.278483992467043e-05,
+      "loss": 0.357,
+      "step": 3115
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.278248587570622e-05,
+      "loss": 0.3223,
+      "step": 3116
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2780131826742e-05,
+      "loss": 0.2803,
+      "step": 3117
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.277777777777778e-05,
+      "loss": 0.4271,
+      "step": 3118
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.277542372881356e-05,
+      "loss": 0.3031,
+      "step": 3119
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2773069679849346e-05,
+      "loss": 0.517,
+      "step": 3120
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.277071563088512e-05,
+      "loss": 0.4653,
+      "step": 3121
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.276836158192091e-05,
+      "loss": 0.3251,
+      "step": 3122
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2766007532956684e-05,
+      "loss": 0.2874,
+      "step": 3123
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2763653483992475e-05,
+      "loss": 0.2918,
+      "step": 3124
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.276129943502825e-05,
+      "loss": 0.4091,
+      "step": 3125
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.275894538606403e-05,
+      "loss": 0.2631,
+      "step": 3126
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.275659133709981e-05,
+      "loss": 0.3513,
+      "step": 3127
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.27542372881356e-05,
+      "loss": 0.3402,
+      "step": 3128
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2751883239171374e-05,
+      "loss": 0.2918,
+      "step": 3129
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.274952919020716e-05,
+      "loss": 0.3449,
+      "step": 3130
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.274717514124294e-05,
+      "loss": 0.2857,
+      "step": 3131
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.274482109227872e-05,
+      "loss": 0.4675,
+      "step": 3132
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.27424670433145e-05,
+      "loss": 0.2987,
+      "step": 3133
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.274011299435029e-05,
+      "loss": 0.418,
+      "step": 3134
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2737758945386064e-05,
+      "loss": 0.3257,
+      "step": 3135
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.273540489642185e-05,
+      "loss": 0.4955,
+      "step": 3136
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.273305084745763e-05,
+      "loss": 0.471,
+      "step": 3137
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.273069679849341e-05,
+      "loss": 0.375,
+      "step": 3138
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.272834274952919e-05,
+      "loss": 0.4287,
+      "step": 3139
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.272598870056497e-05,
+      "loss": 0.3568,
+      "step": 3140
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.272363465160076e-05,
+      "loss": 0.4804,
+      "step": 3141
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.272128060263654e-05,
+      "loss": 0.3262,
+      "step": 3142
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2718926553672315e-05,
+      "loss": 0.3214,
+      "step": 3143
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2716572504708106e-05,
+      "loss": 0.4031,
+      "step": 3144
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.271421845574388e-05,
+      "loss": 0.263,
+      "step": 3145
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.271186440677966e-05,
+      "loss": 0.3872,
+      "step": 3146
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2709510357815444e-05,
+      "loss": 0.2715,
+      "step": 3147
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.270715630885123e-05,
+      "loss": 0.401,
+      "step": 3148
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2704802259887005e-05,
+      "loss": 0.2682,
+      "step": 3149
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.270244821092279e-05,
+      "loss": 0.3939,
+      "step": 3150
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.270009416195857e-05,
+      "loss": 0.4507,
+      "step": 3151
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.269774011299435e-05,
+      "loss": 0.4887,
+      "step": 3152
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2695386064030134e-05,
+      "loss": 0.3405,
+      "step": 3153
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.269303201506592e-05,
+      "loss": 0.3536,
+      "step": 3154
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2690677966101695e-05,
+      "loss": 0.3464,
+      "step": 3155
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.268832391713748e-05,
+      "loss": 0.2618,
+      "step": 3156
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2685969868173256e-05,
+      "loss": 0.2772,
+      "step": 3157
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.268361581920905e-05,
+      "loss": 0.511,
+      "step": 3158
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2681261770244824e-05,
+      "loss": 0.2972,
+      "step": 3159
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.26789077212806e-05,
+      "loss": 0.261,
+      "step": 3160
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.267655367231639e-05,
+      "loss": 0.3146,
+      "step": 3161
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.267419962335217e-05,
+      "loss": 0.3121,
+      "step": 3162
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.2671845574387946e-05,
+      "loss": 0.4454,
+      "step": 3163
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.266949152542373e-05,
+      "loss": 0.5027,
+      "step": 3164
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.2667137476459514e-05,
+      "loss": 0.4263,
+      "step": 3165
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.266478342749529e-05,
+      "loss": 0.377,
+      "step": 3166
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.2662429378531075e-05,
+      "loss": 0.3492,
+      "step": 3167
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.266007532956686e-05,
+      "loss": 0.3601,
+      "step": 3168
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.2657721280602636e-05,
+      "loss": 0.4336,
+      "step": 3169
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.265536723163842e-05,
+      "loss": 0.2876,
+      "step": 3170
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.2653013182674204e-05,
+      "loss": 0.4479,
+      "step": 3171
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.265065913370999e-05,
+      "loss": 0.4789,
+      "step": 3172
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.2648305084745765e-05,
+      "loss": 0.5306,
+      "step": 3173
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.264595103578154e-05,
+      "loss": 0.3822,
+      "step": 3174
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.264359698681733e-05,
+      "loss": 0.2929,
+      "step": 3175
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.264124293785311e-05,
+      "loss": 0.3629,
+      "step": 3176
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.263888888888889e-05,
+      "loss": 0.3189,
+      "step": 3177
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.263653483992468e-05,
+      "loss": 0.2494,
+      "step": 3178
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.2634180790960455e-05,
+      "loss": 0.3647,
+      "step": 3179
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.263182674199623e-05,
+      "loss": 0.3084,
+      "step": 3180
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.2629472693032016e-05,
+      "loss": 0.3602,
+      "step": 3181
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.26271186440678e-05,
+      "loss": 0.2997,
+      "step": 3182
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.262476459510358e-05,
+      "loss": 0.3563,
+      "step": 3183
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.262241054613936e-05,
+      "loss": 0.2267,
+      "step": 3184
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.2620056497175145e-05,
+      "loss": 0.4389,
+      "step": 3185
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.261770244821092e-05,
+      "loss": 0.4264,
+      "step": 3186
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.2615348399246706e-05,
+      "loss": 0.3153,
+      "step": 3187
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.261299435028249e-05,
+      "loss": 0.3622,
+      "step": 3188
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.2610640301318274e-05,
+      "loss": 0.384,
+      "step": 3189
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.260828625235405e-05,
+      "loss": 0.4389,
+      "step": 3190
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.260593220338983e-05,
+      "loss": 0.2393,
+      "step": 3191
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.260357815442562e-05,
+      "loss": 0.4048,
+      "step": 3192
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.2601224105461396e-05,
+      "loss": 0.3434,
+      "step": 3193
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.259887005649717e-05,
+      "loss": 0.2772,
+      "step": 3194
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.2596516007532964e-05,
+      "loss": 0.3518,
+      "step": 3195
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.259416195856874e-05,
+      "loss": 0.4057,
+      "step": 3196
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.259180790960452e-05,
+      "loss": 0.2676,
+      "step": 3197
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.25894538606403e-05,
+      "loss": 0.593,
+      "step": 3198
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.2587099811676086e-05,
+      "loss": 0.3141,
+      "step": 3199
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.258474576271186e-05,
+      "loss": 0.4511,
+      "step": 3200
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.258239171374765e-05,
+      "loss": 0.2833,
+      "step": 3201
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.258003766478343e-05,
+      "loss": 0.4353,
+      "step": 3202
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.2577683615819215e-05,
+      "loss": 0.3256,
+      "step": 3203
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.257532956685499e-05,
+      "loss": 0.3297,
+      "step": 3204
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2572975517890776e-05,
+      "loss": 0.3062,
+      "step": 3205
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.257062146892656e-05,
+      "loss": 0.3118,
+      "step": 3206
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.256826741996234e-05,
+      "loss": 0.3298,
+      "step": 3207
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2565913370998114e-05,
+      "loss": 0.2962,
+      "step": 3208
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2563559322033905e-05,
+      "loss": 0.2894,
+      "step": 3209
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.256120527306968e-05,
+      "loss": 0.4105,
+      "step": 3210
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.255885122410546e-05,
+      "loss": 0.4774,
+      "step": 3211
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.255649717514125e-05,
+      "loss": 0.4046,
+      "step": 3212
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.255414312617703e-05,
+      "loss": 0.674,
+      "step": 3213
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2551789077212804e-05,
+      "loss": 0.4614,
+      "step": 3214
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.254943502824859e-05,
+      "loss": 0.4137,
+      "step": 3215
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.254708097928437e-05,
+      "loss": 0.4459,
+      "step": 3216
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.254472693032015e-05,
+      "loss": 0.5755,
+      "step": 3217
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.254237288135593e-05,
+      "loss": 0.347,
+      "step": 3218
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.254001883239172e-05,
+      "loss": 0.3965,
+      "step": 3219
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.25376647834275e-05,
+      "loss": 0.4126,
+      "step": 3220
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.253531073446328e-05,
+      "loss": 0.3681,
+      "step": 3221
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.253295668549906e-05,
+      "loss": 0.3765,
+      "step": 3222
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2530602636534846e-05,
+      "loss": 0.3049,
+      "step": 3223
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.252824858757062e-05,
+      "loss": 0.3141,
+      "step": 3224
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.25258945386064e-05,
+      "loss": 0.4961,
+      "step": 3225
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.252354048964219e-05,
+      "loss": 0.3519,
+      "step": 3226
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.252118644067797e-05,
+      "loss": 0.287,
+      "step": 3227
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2518832391713745e-05,
+      "loss": 0.2471,
+      "step": 3228
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2516478342749536e-05,
+      "loss": 0.3591,
+      "step": 3229
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.251412429378531e-05,
+      "loss": 0.4715,
+      "step": 3230
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.251177024482109e-05,
+      "loss": 0.4504,
+      "step": 3231
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2509416195856874e-05,
+      "loss": 0.3054,
+      "step": 3232
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.250706214689266e-05,
+      "loss": 0.3547,
+      "step": 3233
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2504708097928435e-05,
+      "loss": 0.3656,
+      "step": 3234
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.250235404896422e-05,
+      "loss": 0.4668,
+      "step": 3235
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.25e-05,
+      "loss": 0.3038,
+      "step": 3236
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.249764595103579e-05,
+      "loss": 0.4148,
+      "step": 3237
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2495291902071564e-05,
+      "loss": 0.4396,
+      "step": 3238
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.249293785310735e-05,
+      "loss": 0.4057,
+      "step": 3239
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.249058380414313e-05,
+      "loss": 0.2989,
+      "step": 3240
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.248822975517891e-05,
+      "loss": 0.3261,
+      "step": 3241
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2485875706214686e-05,
+      "loss": 0.4206,
+      "step": 3242
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.248352165725048e-05,
+      "loss": 0.3372,
+      "step": 3243
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2481167608286254e-05,
+      "loss": 0.3621,
+      "step": 3244
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.247881355932203e-05,
+      "loss": 0.5025,
+      "step": 3245
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.247645951035782e-05,
+      "loss": 0.3611,
+      "step": 3246
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.24741054613936e-05,
+      "loss": 0.4914,
+      "step": 3247
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2471751412429376e-05,
+      "loss": 0.4629,
+      "step": 3248
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.246939736346516e-05,
+      "loss": 0.3509,
+      "step": 3249
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2467043314500944e-05,
+      "loss": 0.2644,
+      "step": 3250
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.246468926553673e-05,
+      "loss": 0.3423,
+      "step": 3251
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2462335216572505e-05,
+      "loss": 0.3505,
+      "step": 3252
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.245998116760829e-05,
+      "loss": 0.3892,
+      "step": 3253
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.245762711864407e-05,
+      "loss": 0.3185,
+      "step": 3254
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.245527306967985e-05,
+      "loss": 0.3714,
+      "step": 3255
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2452919020715634e-05,
+      "loss": 0.3998,
+      "step": 3256
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.245056497175142e-05,
+      "loss": 0.5582,
+      "step": 3257
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2448210922787195e-05,
+      "loss": 0.2464,
+      "step": 3258
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.244585687382297e-05,
+      "loss": 0.3095,
+      "step": 3259
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.244350282485876e-05,
+      "loss": 0.465,
+      "step": 3260
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.244114877589454e-05,
+      "loss": 0.2385,
+      "step": 3261
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.243879472693032e-05,
+      "loss": 0.3075,
+      "step": 3262
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.243644067796611e-05,
+      "loss": 0.3798,
+      "step": 3263
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2434086629001885e-05,
+      "loss": 0.4469,
+      "step": 3264
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.243173258003766e-05,
+      "loss": 0.4539,
+      "step": 3265
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2429378531073446e-05,
+      "loss": 0.249,
+      "step": 3266
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.242702448210923e-05,
+      "loss": 0.4465,
+      "step": 3267
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2424670433145014e-05,
+      "loss": 0.2952,
+      "step": 3268
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.242231638418079e-05,
+      "loss": 0.3636,
+      "step": 3269
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2419962335216575e-05,
+      "loss": 0.2925,
+      "step": 3270
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.241760828625236e-05,
+      "loss": 0.4174,
+      "step": 3271
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2415254237288136e-05,
+      "loss": 0.3463,
+      "step": 3272
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.241290018832392e-05,
+      "loss": 0.3204,
+      "step": 3273
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2410546139359704e-05,
+      "loss": 0.3153,
+      "step": 3274
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.240819209039548e-05,
+      "loss": 0.3249,
+      "step": 3275
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.240583804143126e-05,
+      "loss": 0.4146,
+      "step": 3276
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.240348399246705e-05,
+      "loss": 0.3472,
+      "step": 3277
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2401129943502826e-05,
+      "loss": 0.3696,
+      "step": 3278
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.23987758945386e-05,
+      "loss": 0.4529,
+      "step": 3279
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2396421845574394e-05,
+      "loss": 0.2662,
+      "step": 3280
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.239406779661017e-05,
+      "loss": 0.3155,
+      "step": 3281
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2391713747645955e-05,
+      "loss": 0.4728,
+      "step": 3282
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.238935969868173e-05,
+      "loss": 0.2821,
+      "step": 3283
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2387005649717516e-05,
+      "loss": 0.248,
+      "step": 3284
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.23846516007533e-05,
+      "loss": 0.3099,
+      "step": 3285
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.238229755178908e-05,
+      "loss": 0.4452,
+      "step": 3286
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.237994350282486e-05,
+      "loss": 0.3915,
+      "step": 3287
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2377589453860645e-05,
+      "loss": 0.5636,
+      "step": 3288
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.237523540489642e-05,
+      "loss": 0.2301,
+      "step": 3289
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2372881355932206e-05,
+      "loss": 0.3036,
+      "step": 3290
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.237052730696799e-05,
+      "loss": 0.3961,
+      "step": 3291
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.236817325800377e-05,
+      "loss": 0.2977,
+      "step": 3292
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2365819209039544e-05,
+      "loss": 0.3239,
+      "step": 3293
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2363465160075335e-05,
+      "loss": 0.3532,
+      "step": 3294
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.236111111111111e-05,
+      "loss": 0.3458,
+      "step": 3295
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.235875706214689e-05,
+      "loss": 0.3732,
+      "step": 3296
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.235640301318268e-05,
+      "loss": 0.3505,
+      "step": 3297
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.235404896421846e-05,
+      "loss": 0.2934,
+      "step": 3298
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.235169491525424e-05,
+      "loss": 0.49,
+      "step": 3299
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.234934086629002e-05,
+      "loss": 0.3293,
+      "step": 3300
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.23469868173258e-05,
+      "loss": 0.3554,
+      "step": 3301
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2344632768361586e-05,
+      "loss": 0.2945,
+      "step": 3302
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.234227871939736e-05,
+      "loss": 0.2878,
+      "step": 3303
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.233992467043315e-05,
+      "loss": 0.3584,
+      "step": 3304
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.233757062146893e-05,
+      "loss": 0.3281,
+      "step": 3305
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.233521657250471e-05,
+      "loss": 0.3832,
+      "step": 3306
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.233286252354049e-05,
+      "loss": 0.404,
+      "step": 3307
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2330508474576276e-05,
+      "loss": 0.4399,
+      "step": 3308
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.232815442561205e-05,
+      "loss": 0.3027,
+      "step": 3309
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.232580037664784e-05,
+      "loss": 0.4054,
+      "step": 3310
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.232344632768362e-05,
+      "loss": 0.2449,
+      "step": 3311
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.23210922787194e-05,
+      "loss": 0.3089,
+      "step": 3312
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2318738229755176e-05,
+      "loss": 0.3358,
+      "step": 3313
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2316384180790966e-05,
+      "loss": 0.4598,
+      "step": 3314
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.231403013182674e-05,
+      "loss": 0.3262,
+      "step": 3315
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.231167608286253e-05,
+      "loss": 0.3449,
+      "step": 3316
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2309322033898304e-05,
+      "loss": 0.4095,
+      "step": 3317
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.230696798493409e-05,
+      "loss": 0.2389,
+      "step": 3318
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.230461393596987e-05,
+      "loss": 0.3606,
+      "step": 3319
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.230225988700565e-05,
+      "loss": 0.3959,
+      "step": 3320
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.229990583804143e-05,
+      "loss": 0.4018,
+      "step": 3321
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.229755178907722e-05,
+      "loss": 0.3394,
+      "step": 3322
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2295197740112994e-05,
+      "loss": 0.4139,
+      "step": 3323
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.229284369114878e-05,
+      "loss": 0.2225,
+      "step": 3324
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.229048964218456e-05,
+      "loss": 0.3729,
+      "step": 3325
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.228813559322034e-05,
+      "loss": 0.3831,
+      "step": 3326
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.228578154425612e-05,
+      "loss": 0.4162,
+      "step": 3327
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.228342749529191e-05,
+      "loss": 0.3911,
+      "step": 3328
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2281073446327684e-05,
+      "loss": 0.2972,
+      "step": 3329
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.227871939736347e-05,
+      "loss": 0.2615,
+      "step": 3330
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.227636534839925e-05,
+      "loss": 0.4228,
+      "step": 3331
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.227401129943503e-05,
+      "loss": 0.4646,
+      "step": 3332
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.227165725047081e-05,
+      "loss": 0.3643,
+      "step": 3333
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.226930320150659e-05,
+      "loss": 0.3246,
+      "step": 3334
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2266949152542374e-05,
+      "loss": 0.3838,
+      "step": 3335
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.226459510357816e-05,
+      "loss": 0.4333,
+      "step": 3336
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2262241054613935e-05,
+      "loss": 0.3168,
+      "step": 3337
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.225988700564972e-05,
+      "loss": 0.2584,
+      "step": 3338
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.22575329566855e-05,
+      "loss": 0.3375,
+      "step": 3339
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.225517890772128e-05,
+      "loss": 0.3049,
+      "step": 3340
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2252824858757064e-05,
+      "loss": 0.3487,
+      "step": 3341
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.225047080979285e-05,
+      "loss": 0.4672,
+      "step": 3342
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2248116760828625e-05,
+      "loss": 0.4066,
+      "step": 3343
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.224576271186441e-05,
+      "loss": 0.3666,
+      "step": 3344
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.224340866290019e-05,
+      "loss": 0.3773,
+      "step": 3345
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.224105461393597e-05,
+      "loss": 0.4029,
+      "step": 3346
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2238700564971754e-05,
+      "loss": 0.2735,
+      "step": 3347
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.223634651600754e-05,
+      "loss": 0.3094,
+      "step": 3348
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2233992467043315e-05,
+      "loss": 0.4059,
+      "step": 3349
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.22316384180791e-05,
+      "loss": 0.363,
+      "step": 3350
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2229284369114877e-05,
+      "loss": 0.2973,
+      "step": 3351
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.222693032015066e-05,
+      "loss": 0.3273,
+      "step": 3352
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2224576271186444e-05,
+      "loss": 0.4019,
+      "step": 3353
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.222222222222222e-05,
+      "loss": 0.4563,
+      "step": 3354
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2219868173258005e-05,
+      "loss": 0.4432,
+      "step": 3355
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.221751412429379e-05,
+      "loss": 0.4725,
+      "step": 3356
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2215160075329567e-05,
+      "loss": 0.4327,
+      "step": 3357
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.221280602636535e-05,
+      "loss": 0.3052,
+      "step": 3358
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2210451977401134e-05,
+      "loss": 0.3011,
+      "step": 3359
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.220809792843691e-05,
+      "loss": 0.3116,
+      "step": 3360
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2205743879472695e-05,
+      "loss": 0.274,
+      "step": 3361
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.220338983050848e-05,
+      "loss": 0.3576,
+      "step": 3362
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2201035781544257e-05,
+      "loss": 0.2957,
+      "step": 3363
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.219868173258004e-05,
+      "loss": 0.2739,
+      "step": 3364
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2196327683615824e-05,
+      "loss": 0.2917,
+      "step": 3365
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.21939736346516e-05,
+      "loss": 0.3055,
+      "step": 3366
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2191619585687385e-05,
+      "loss": 0.5123,
+      "step": 3367
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.218926553672316e-05,
+      "loss": 0.5251,
+      "step": 3368
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2186911487758947e-05,
+      "loss": 0.317,
+      "step": 3369
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.218455743879473e-05,
+      "loss": 0.179,
+      "step": 3370
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.218220338983051e-05,
+      "loss": 0.4097,
+      "step": 3371
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.217984934086629e-05,
+      "loss": 0.3635,
+      "step": 3372
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2177495291902075e-05,
+      "loss": 0.3032,
+      "step": 3373
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.217514124293785e-05,
+      "loss": 0.3497,
+      "step": 3374
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2172787193973637e-05,
+      "loss": 0.321,
+      "step": 3375
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.217043314500942e-05,
+      "loss": 0.3413,
+      "step": 3376
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.21680790960452e-05,
+      "loss": 0.3757,
+      "step": 3377
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.216572504708098e-05,
+      "loss": 0.2995,
+      "step": 3378
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2163370998116765e-05,
+      "loss": 0.2707,
+      "step": 3379
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.216101694915254e-05,
+      "loss": 0.3229,
+      "step": 3380
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2158662900188327e-05,
+      "loss": 0.3733,
+      "step": 3381
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.215630885122411e-05,
+      "loss": 0.441,
+      "step": 3382
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.215395480225989e-05,
+      "loss": 0.2912,
+      "step": 3383
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.215160075329567e-05,
+      "loss": 0.3885,
+      "step": 3384
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2149246704331455e-05,
+      "loss": 0.2715,
+      "step": 3385
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.214689265536723e-05,
+      "loss": 0.4382,
+      "step": 3386
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2144538606403017e-05,
+      "loss": 0.3133,
+      "step": 3387
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2142184557438794e-05,
+      "loss": 0.4074,
+      "step": 3388
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.213983050847458e-05,
+      "loss": 0.3254,
+      "step": 3389
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.213747645951036e-05,
+      "loss": 0.39,
+      "step": 3390
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.213512241054614e-05,
+      "loss": 0.3868,
+      "step": 3391
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.213276836158192e-05,
+      "loss": 0.3381,
+      "step": 3392
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2130414312617706e-05,
+      "loss": 0.6521,
+      "step": 3393
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2128060263653484e-05,
+      "loss": 0.322,
+      "step": 3394
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.212570621468927e-05,
+      "loss": 0.4645,
+      "step": 3395
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.212335216572505e-05,
+      "loss": 0.2788,
+      "step": 3396
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.212099811676083e-05,
+      "loss": 0.2932,
+      "step": 3397
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.211864406779661e-05,
+      "loss": 0.3572,
+      "step": 3398
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2116290018832396e-05,
+      "loss": 0.3381,
+      "step": 3399
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2113935969868174e-05,
+      "loss": 0.344,
+      "step": 3400
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.211158192090396e-05,
+      "loss": 0.2509,
+      "step": 3401
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.210922787193974e-05,
+      "loss": 0.295,
+      "step": 3402
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.210687382297552e-05,
+      "loss": 0.2841,
+      "step": 3403
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.21045197740113e-05,
+      "loss": 0.3192,
+      "step": 3404
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.210216572504708e-05,
+      "loss": 0.2735,
+      "step": 3405
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2099811676082864e-05,
+      "loss": 0.5026,
+      "step": 3406
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.209745762711865e-05,
+      "loss": 0.2736,
+      "step": 3407
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2095103578154425e-05,
+      "loss": 0.2409,
+      "step": 3408
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.209274952919021e-05,
+      "loss": 0.3775,
+      "step": 3409
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.209039548022599e-05,
+      "loss": 0.3507,
+      "step": 3410
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.208804143126177e-05,
+      "loss": 0.3945,
+      "step": 3411
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2085687382297554e-05,
+      "loss": 0.3723,
+      "step": 3412
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.208333333333334e-05,
+      "loss": 0.3434,
+      "step": 3413
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2080979284369115e-05,
+      "loss": 0.4315,
+      "step": 3414
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.20786252354049e-05,
+      "loss": 0.4015,
+      "step": 3415
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.207627118644068e-05,
+      "loss": 0.4905,
+      "step": 3416
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.207391713747646e-05,
+      "loss": 0.2975,
+      "step": 3417
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.2071563088512244e-05,
+      "loss": 0.4449,
+      "step": 3418
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.206920903954803e-05,
+      "loss": 0.2685,
+      "step": 3419
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.2066854990583805e-05,
+      "loss": 0.4715,
+      "step": 3420
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.206450094161959e-05,
+      "loss": 0.3588,
+      "step": 3421
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.2062146892655366e-05,
+      "loss": 0.3287,
+      "step": 3422
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.205979284369115e-05,
+      "loss": 0.39,
+      "step": 3423
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.2057438794726934e-05,
+      "loss": 0.4097,
+      "step": 3424
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.205508474576271e-05,
+      "loss": 0.2905,
+      "step": 3425
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.2052730696798495e-05,
+      "loss": 0.2147,
+      "step": 3426
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.205037664783428e-05,
+      "loss": 0.2873,
+      "step": 3427
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.2048022598870056e-05,
+      "loss": 0.4103,
+      "step": 3428
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.204566854990584e-05,
+      "loss": 0.4219,
+      "step": 3429
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.2043314500941624e-05,
+      "loss": 0.2975,
+      "step": 3430
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.20409604519774e-05,
+      "loss": 0.3542,
+      "step": 3431
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.2038606403013185e-05,
+      "loss": 0.4903,
+      "step": 3432
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.203625235404897e-05,
+      "loss": 0.2717,
+      "step": 3433
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.2033898305084746e-05,
+      "loss": 0.3295,
+      "step": 3434
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.203154425612053e-05,
+      "loss": 0.2828,
+      "step": 3435
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.2029190207156314e-05,
+      "loss": 0.3796,
+      "step": 3436
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.202683615819209e-05,
+      "loss": 0.4208,
+      "step": 3437
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.2024482109227875e-05,
+      "loss": 0.3022,
+      "step": 3438
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.202212806026365e-05,
+      "loss": 0.4546,
+      "step": 3439
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.201977401129944e-05,
+      "loss": 0.445,
+      "step": 3440
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.201741996233522e-05,
+      "loss": 0.4128,
+      "step": 3441
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.2015065913371e-05,
+      "loss": 0.3269,
+      "step": 3442
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.201271186440678e-05,
+      "loss": 0.3603,
+      "step": 3443
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.2010357815442565e-05,
+      "loss": 0.4121,
+      "step": 3444
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.200800376647834e-05,
+      "loss": 0.4098,
+      "step": 3445
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.2005649717514126e-05,
+      "loss": 0.4213,
+      "step": 3446
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.200329566854991e-05,
+      "loss": 0.2908,
+      "step": 3447
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.200094161958569e-05,
+      "loss": 0.3033,
+      "step": 3448
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.199858757062147e-05,
+      "loss": 0.3265,
+      "step": 3449
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.1996233521657255e-05,
+      "loss": 0.3366,
+      "step": 3450
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.199387947269303e-05,
+      "loss": 0.3231,
+      "step": 3451
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.1991525423728816e-05,
+      "loss": 0.3705,
+      "step": 3452
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.19891713747646e-05,
+      "loss": 0.3081,
+      "step": 3453
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.198681732580038e-05,
+      "loss": 0.3322,
+      "step": 3454
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.198446327683616e-05,
+      "loss": 0.2886,
+      "step": 3455
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.198210922787194e-05,
+      "loss": 0.3926,
+      "step": 3456
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.197975517890773e-05,
+      "loss": 0.2733,
+      "step": 3457
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.1977401129943506e-05,
+      "loss": 0.358,
+      "step": 3458
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.197504708097928e-05,
+      "loss": 0.336,
+      "step": 3459
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.1972693032015074e-05,
+      "loss": 0.3106,
+      "step": 3460
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.197033898305085e-05,
+      "loss": 0.3416,
+      "step": 3461
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.196798493408663e-05,
+      "loss": 0.3755,
+      "step": 3462
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.196563088512241e-05,
+      "loss": 0.2685,
+      "step": 3463
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.1963276836158196e-05,
+      "loss": 0.4633,
+      "step": 3464
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.196092278719397e-05,
+      "loss": 0.2442,
+      "step": 3465
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.195856873822976e-05,
+      "loss": 0.2893,
+      "step": 3466
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.195621468926554e-05,
+      "loss": 0.4047,
+      "step": 3467
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.195386064030132e-05,
+      "loss": 0.2612,
+      "step": 3468
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.19515065913371e-05,
+      "loss": 0.379,
+      "step": 3469
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.1949152542372886e-05,
+      "loss": 0.3659,
+      "step": 3470
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.194679849340866e-05,
+      "loss": 0.254,
+      "step": 3471
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.194444444444445e-05,
+      "loss": 0.4041,
+      "step": 3472
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.1942090395480224e-05,
+      "loss": 0.5208,
+      "step": 3473
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.1939736346516015e-05,
+      "loss": 0.4372,
+      "step": 3474
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.193738229755179e-05,
+      "loss": 0.3101,
+      "step": 3475
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.193502824858757e-05,
+      "loss": 0.4078,
+      "step": 3476
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.193267419962336e-05,
+      "loss": 0.3296,
+      "step": 3477
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.193032015065914e-05,
+      "loss": 0.4229,
+      "step": 3478
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.1927966101694914e-05,
+      "loss": 0.3967,
+      "step": 3479
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.19256120527307e-05,
+      "loss": 0.3783,
+      "step": 3480
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.192325800376648e-05,
+      "loss": 0.3343,
+      "step": 3481
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.192090395480226e-05,
+      "loss": 0.3151,
+      "step": 3482
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.191854990583804e-05,
+      "loss": 0.4266,
+      "step": 3483
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.191619585687383e-05,
+      "loss": 0.3678,
+      "step": 3484
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1913841807909604e-05,
+      "loss": 0.3794,
+      "step": 3485
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.191148775894539e-05,
+      "loss": 0.3685,
+      "step": 3486
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.190913370998117e-05,
+      "loss": 0.3165,
+      "step": 3487
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1906779661016956e-05,
+      "loss": 0.2414,
+      "step": 3488
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.190442561205273e-05,
+      "loss": 0.3447,
+      "step": 3489
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.190207156308851e-05,
+      "loss": 0.3056,
+      "step": 3490
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.18997175141243e-05,
+      "loss": 0.31,
+      "step": 3491
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.189736346516008e-05,
+      "loss": 0.1767,
+      "step": 3492
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1895009416195855e-05,
+      "loss": 0.2968,
+      "step": 3493
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1892655367231646e-05,
+      "loss": 0.3141,
+      "step": 3494
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.189030131826742e-05,
+      "loss": 0.4384,
+      "step": 3495
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.18879472693032e-05,
+      "loss": 0.2893,
+      "step": 3496
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1885593220338984e-05,
+      "loss": 0.3286,
+      "step": 3497
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.188323917137477e-05,
+      "loss": 0.3504,
+      "step": 3498
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1880885122410545e-05,
+      "loss": 0.3075,
+      "step": 3499
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.187853107344633e-05,
+      "loss": 0.3887,
+      "step": 3500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.187617702448211e-05,
+      "loss": 0.2818,
+      "step": 3501
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.187382297551789e-05,
+      "loss": 0.3308,
+      "step": 3502
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.1871468926553674e-05,
+      "loss": 0.4911,
+      "step": 3503
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.186911487758946e-05,
+      "loss": 0.3996,
+      "step": 3504
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.186676082862524e-05,
+      "loss": 0.4467,
+      "step": 3505
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.186440677966102e-05,
+      "loss": 0.2695,
+      "step": 3506
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.1862052730696796e-05,
+      "loss": 0.3275,
+      "step": 3507
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.185969868173259e-05,
+      "loss": 0.3268,
+      "step": 3508
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.1857344632768364e-05,
+      "loss": 0.3121,
+      "step": 3509
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.185499058380414e-05,
+      "loss": 0.3408,
+      "step": 3510
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.185263653483993e-05,
+      "loss": 0.2405,
+      "step": 3511
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.185028248587571e-05,
+      "loss": 0.2813,
+      "step": 3512
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.1847928436911486e-05,
+      "loss": 0.3204,
+      "step": 3513
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.184557438794727e-05,
+      "loss": 0.5245,
+      "step": 3514
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.1843220338983054e-05,
+      "loss": 0.4202,
+      "step": 3515
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.184086629001883e-05,
+      "loss": 0.2608,
+      "step": 3516
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.1838512241054615e-05,
+      "loss": 0.3965,
+      "step": 3517
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.18361581920904e-05,
+      "loss": 0.3328,
+      "step": 3518
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.183380414312618e-05,
+      "loss": 0.2978,
+      "step": 3519
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.183145009416196e-05,
+      "loss": 0.3512,
+      "step": 3520
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.1829096045197744e-05,
+      "loss": 0.2921,
+      "step": 3521
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.182674199623353e-05,
+      "loss": 0.3241,
+      "step": 3522
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.1824387947269305e-05,
+      "loss": 0.4209,
+      "step": 3523
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.182203389830508e-05,
+      "loss": 0.3281,
+      "step": 3524
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.181967984934087e-05,
+      "loss": 0.2821,
+      "step": 3525
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.181732580037665e-05,
+      "loss": 0.4379,
+      "step": 3526
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.181497175141243e-05,
+      "loss": 0.4275,
+      "step": 3527
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.181261770244822e-05,
+      "loss": 0.4434,
+      "step": 3528
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1810263653483995e-05,
+      "loss": 0.4084,
+      "step": 3529
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.180790960451977e-05,
+      "loss": 0.5475,
+      "step": 3530
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1805555555555556e-05,
+      "loss": 0.2645,
+      "step": 3531
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.180320150659134e-05,
+      "loss": 0.3739,
+      "step": 3532
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.180084745762712e-05,
+      "loss": 0.3264,
+      "step": 3533
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.17984934086629e-05,
+      "loss": 0.3516,
+      "step": 3534
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1796139359698685e-05,
+      "loss": 0.3307,
+      "step": 3535
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.179378531073447e-05,
+      "loss": 0.3057,
+      "step": 3536
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1791431261770246e-05,
+      "loss": 0.2987,
+      "step": 3537
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.178907721280603e-05,
+      "loss": 0.3329,
+      "step": 3538
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1786723163841814e-05,
+      "loss": 0.412,
+      "step": 3539
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.178436911487759e-05,
+      "loss": 0.4517,
+      "step": 3540
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.178201506591337e-05,
+      "loss": 0.366,
+      "step": 3541
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.177966101694916e-05,
+      "loss": 0.4419,
+      "step": 3542
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1777306967984936e-05,
+      "loss": 0.2259,
+      "step": 3543
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.177495291902071e-05,
+      "loss": 0.26,
+      "step": 3544
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1772598870056504e-05,
+      "loss": 0.3708,
+      "step": 3545
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.177024482109228e-05,
+      "loss": 0.4396,
+      "step": 3546
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.176789077212806e-05,
+      "loss": 0.4213,
+      "step": 3547
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.176553672316384e-05,
+      "loss": 0.3226,
+      "step": 3548
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1763182674199626e-05,
+      "loss": 0.256,
+      "step": 3549
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.17608286252354e-05,
+      "loss": 0.3604,
+      "step": 3550
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.175847457627119e-05,
+      "loss": 0.2419,
+      "step": 3551
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.175612052730697e-05,
+      "loss": 0.4155,
+      "step": 3552
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1753766478342755e-05,
+      "loss": 0.3894,
+      "step": 3553
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.175141242937853e-05,
+      "loss": 0.3773,
+      "step": 3554
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1749058380414316e-05,
+      "loss": 0.251,
+      "step": 3555
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.17467043314501e-05,
+      "loss": 0.4264,
+      "step": 3556
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.174435028248588e-05,
+      "loss": 0.3409,
+      "step": 3557
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1741996233521654e-05,
+      "loss": 0.3461,
+      "step": 3558
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1739642184557445e-05,
+      "loss": 0.2608,
+      "step": 3559
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.173728813559322e-05,
+      "loss": 0.4705,
+      "step": 3560
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1734934086629e-05,
+      "loss": 0.3048,
+      "step": 3561
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.173258003766479e-05,
+      "loss": 0.3204,
+      "step": 3562
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.173022598870057e-05,
+      "loss": 0.3213,
+      "step": 3563
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1727871939736344e-05,
+      "loss": 0.4579,
+      "step": 3564
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.172551789077213e-05,
+      "loss": 0.3893,
+      "step": 3565
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.172316384180791e-05,
+      "loss": 0.3343,
+      "step": 3566
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.1720809792843696e-05,
+      "loss": 0.3733,
+      "step": 3567
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.171845574387947e-05,
+      "loss": 0.3359,
+      "step": 3568
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.171610169491526e-05,
+      "loss": 0.3532,
+      "step": 3569
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.171374764595104e-05,
+      "loss": 0.3578,
+      "step": 3570
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.171139359698682e-05,
+      "loss": 0.3985,
+      "step": 3571
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.17090395480226e-05,
+      "loss": 0.3282,
+      "step": 3572
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.1706685499058386e-05,
+      "loss": 0.4852,
+      "step": 3573
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.170433145009416e-05,
+      "loss": 0.3116,
+      "step": 3574
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.170197740112994e-05,
+      "loss": 0.3296,
+      "step": 3575
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.169962335216573e-05,
+      "loss": 0.3398,
+      "step": 3576
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.169726930320151e-05,
+      "loss": 0.2752,
+      "step": 3577
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.1694915254237285e-05,
+      "loss": 0.3873,
+      "step": 3578
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.1692561205273076e-05,
+      "loss": 0.2099,
+      "step": 3579
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.169020715630885e-05,
+      "loss": 0.3954,
+      "step": 3580
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.168785310734463e-05,
+      "loss": 0.261,
+      "step": 3581
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.1685499058380414e-05,
+      "loss": 0.3071,
+      "step": 3582
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.16831450094162e-05,
+      "loss": 0.4652,
+      "step": 3583
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.168079096045198e-05,
+      "loss": 0.3058,
+      "step": 3584
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.167843691148776e-05,
+      "loss": 0.3226,
+      "step": 3585
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.167608286252354e-05,
+      "loss": 0.3871,
+      "step": 3586
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.167372881355933e-05,
+      "loss": 0.2255,
+      "step": 3587
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.1671374764595104e-05,
+      "loss": 0.2454,
+      "step": 3588
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.166902071563089e-05,
+      "loss": 0.2206,
+      "step": 3589
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.2466,
+      "step": 3590
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.166431261770245e-05,
+      "loss": 0.4421,
+      "step": 3591
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.1661958568738226e-05,
+      "loss": 0.3376,
+      "step": 3592
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.165960451977402e-05,
+      "loss": 0.4525,
+      "step": 3593
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.1657250470809794e-05,
+      "loss": 0.3043,
+      "step": 3594
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.165489642184557e-05,
+      "loss": 0.4159,
+      "step": 3595
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.165254237288136e-05,
+      "loss": 0.3296,
+      "step": 3596
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.165018832391714e-05,
+      "loss": 0.3585,
+      "step": 3597
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.164783427495292e-05,
+      "loss": 0.2253,
+      "step": 3598
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.16454802259887e-05,
+      "loss": 0.3405,
+      "step": 3599
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.1643126177024484e-05,
+      "loss": 0.4103,
+      "step": 3600
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.164077212806027e-05,
+      "loss": 0.2941,
+      "step": 3601
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.1638418079096045e-05,
+      "loss": 0.4313,
+      "step": 3602
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.163606403013183e-05,
+      "loss": 0.2885,
+      "step": 3603
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.163370998116761e-05,
+      "loss": 0.4736,
+      "step": 3604
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.163135593220339e-05,
+      "loss": 0.321,
+      "step": 3605
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.1629001883239174e-05,
+      "loss": 0.2951,
+      "step": 3606
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.162664783427496e-05,
+      "loss": 0.2732,
+      "step": 3607
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.1624293785310735e-05,
+      "loss": 0.4192,
+      "step": 3608
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.162193973634651e-05,
+      "loss": 0.256,
+      "step": 3609
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.16195856873823e-05,
+      "loss": 0.385,
+      "step": 3610
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.161723163841808e-05,
+      "loss": 0.4116,
+      "step": 3611
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.161487758945386e-05,
+      "loss": 0.2088,
+      "step": 3612
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.161252354048965e-05,
+      "loss": 0.3683,
+      "step": 3613
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1610169491525425e-05,
+      "loss": 0.4134,
+      "step": 3614
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.160781544256121e-05,
+      "loss": 0.2857,
+      "step": 3615
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1605461393596986e-05,
+      "loss": 0.4794,
+      "step": 3616
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.160310734463277e-05,
+      "loss": 0.4198,
+      "step": 3617
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1600753295668554e-05,
+      "loss": 0.3698,
+      "step": 3618
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.159839924670433e-05,
+      "loss": 0.3397,
+      "step": 3619
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1596045197740115e-05,
+      "loss": 0.4524,
+      "step": 3620
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.15936911487759e-05,
+      "loss": 0.258,
+      "step": 3621
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1591337099811676e-05,
+      "loss": 0.3911,
+      "step": 3622
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.158898305084746e-05,
+      "loss": 0.3082,
+      "step": 3623
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1586629001883244e-05,
+      "loss": 0.3703,
+      "step": 3624
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.158427495291902e-05,
+      "loss": 0.3142,
+      "step": 3625
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1581920903954805e-05,
+      "loss": 0.3547,
+      "step": 3626
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.157956685499059e-05,
+      "loss": 0.3614,
+      "step": 3627
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1577212806026366e-05,
+      "loss": 0.3961,
+      "step": 3628
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1574858757062143e-05,
+      "loss": 0.2354,
+      "step": 3629
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.1572504708097934e-05,
+      "loss": 0.3501,
+      "step": 3630
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.157015065913371e-05,
+      "loss": 0.3103,
+      "step": 3631
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.1567796610169495e-05,
+      "loss": 0.3736,
+      "step": 3632
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.156544256120527e-05,
+      "loss": 0.4262,
+      "step": 3633
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.1563088512241056e-05,
+      "loss": 0.3776,
+      "step": 3634
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.156073446327684e-05,
+      "loss": 0.3423,
+      "step": 3635
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.155838041431262e-05,
+      "loss": 0.3781,
+      "step": 3636
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.15560263653484e-05,
+      "loss": 0.3143,
+      "step": 3637
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.1553672316384185e-05,
+      "loss": 0.2675,
+      "step": 3638
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.155131826741996e-05,
+      "loss": 0.2784,
+      "step": 3639
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.1548964218455746e-05,
+      "loss": 0.3319,
+      "step": 3640
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.154661016949153e-05,
+      "loss": 0.2917,
+      "step": 3641
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.154425612052731e-05,
+      "loss": 0.3627,
+      "step": 3642
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.154190207156309e-05,
+      "loss": 0.3191,
+      "step": 3643
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.1539548022598875e-05,
+      "loss": 0.4893,
+      "step": 3644
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.153719397363465e-05,
+      "loss": 0.4374,
+      "step": 3645
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.1534839924670436e-05,
+      "loss": 0.4624,
+      "step": 3646
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.153248587570622e-05,
+      "loss": 0.2273,
+      "step": 3647
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.1530131826742e-05,
+      "loss": 0.3877,
+      "step": 3648
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.152777777777778e-05,
+      "loss": 0.2587,
+      "step": 3649
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.152542372881356e-05,
+      "loss": 0.2417,
+      "step": 3650
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.152306967984934e-05,
+      "loss": 0.4524,
+      "step": 3651
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1520715630885126e-05,
+      "loss": 0.4265,
+      "step": 3652
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1518361581920903e-05,
+      "loss": 0.305,
+      "step": 3653
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.151600753295669e-05,
+      "loss": 0.4563,
+      "step": 3654
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.151365348399247e-05,
+      "loss": 0.385,
+      "step": 3655
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.151129943502825e-05,
+      "loss": 0.303,
+      "step": 3656
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.150894538606403e-05,
+      "loss": 0.3497,
+      "step": 3657
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1506591337099816e-05,
+      "loss": 0.2474,
+      "step": 3658
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1504237288135593e-05,
+      "loss": 0.2029,
+      "step": 3659
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.150188323917138e-05,
+      "loss": 0.249,
+      "step": 3660
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.149952919020716e-05,
+      "loss": 0.4338,
+      "step": 3661
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.149717514124294e-05,
+      "loss": 0.273,
+      "step": 3662
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.149482109227872e-05,
+      "loss": 0.2318,
+      "step": 3663
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1492467043314506e-05,
+      "loss": 0.3579,
+      "step": 3664
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1490112994350283e-05,
+      "loss": 0.2762,
+      "step": 3665
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.148775894538607e-05,
+      "loss": 0.3467,
+      "step": 3666
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1485404896421844e-05,
+      "loss": 0.3095,
+      "step": 3667
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.148305084745763e-05,
+      "loss": 0.4195,
+      "step": 3668
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.148069679849341e-05,
+      "loss": 0.503,
+      "step": 3669
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.147834274952919e-05,
+      "loss": 0.2321,
+      "step": 3670
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1475988700564973e-05,
+      "loss": 0.2527,
+      "step": 3671
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.147363465160076e-05,
+      "loss": 0.2803,
+      "step": 3672
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1471280602636534e-05,
+      "loss": 0.2557,
+      "step": 3673
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.146892655367232e-05,
+      "loss": 0.3734,
+      "step": 3674
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.14665725047081e-05,
+      "loss": 0.2977,
+      "step": 3675
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.146421845574388e-05,
+      "loss": 0.4251,
+      "step": 3676
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1461864406779663e-05,
+      "loss": 0.3963,
+      "step": 3677
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.145951035781545e-05,
+      "loss": 0.4627,
+      "step": 3678
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1457156308851224e-05,
+      "loss": 0.5638,
+      "step": 3679
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.145480225988701e-05,
+      "loss": 0.3628,
+      "step": 3680
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.145244821092279e-05,
+      "loss": 0.3492,
+      "step": 3681
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.145009416195857e-05,
+      "loss": 0.4113,
+      "step": 3682
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1447740112994353e-05,
+      "loss": 0.3381,
+      "step": 3683
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.144538606403013e-05,
+      "loss": 0.5235,
+      "step": 3684
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1443032015065914e-05,
+      "loss": 0.3168,
+      "step": 3685
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.14406779661017e-05,
+      "loss": 0.4324,
+      "step": 3686
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1438323917137476e-05,
+      "loss": 0.5159,
+      "step": 3687
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.143596986817326e-05,
+      "loss": 0.3007,
+      "step": 3688
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1433615819209043e-05,
+      "loss": 0.3441,
+      "step": 3689
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.143126177024482e-05,
+      "loss": 0.2744,
+      "step": 3690
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1428907721280604e-05,
+      "loss": 0.2471,
+      "step": 3691
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.142655367231639e-05,
+      "loss": 0.2224,
+      "step": 3692
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1424199623352166e-05,
+      "loss": 0.3197,
+      "step": 3693
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.142184557438795e-05,
+      "loss": 0.2769,
+      "step": 3694
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1419491525423733e-05,
+      "loss": 0.3861,
+      "step": 3695
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.141713747645951e-05,
+      "loss": 0.2618,
+      "step": 3696
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1414783427495294e-05,
+      "loss": 0.3247,
+      "step": 3697
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.141242937853108e-05,
+      "loss": 0.3266,
+      "step": 3698
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1410075329566856e-05,
+      "loss": 0.4857,
+      "step": 3699
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.140772128060264e-05,
+      "loss": 0.4455,
+      "step": 3700
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1405367231638423e-05,
+      "loss": 0.417,
+      "step": 3701
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.14030131826742e-05,
+      "loss": 0.2848,
+      "step": 3702
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1400659133709984e-05,
+      "loss": 0.3504,
+      "step": 3703
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.139830508474576e-05,
+      "loss": 0.2984,
+      "step": 3704
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1395951035781546e-05,
+      "loss": 0.3203,
+      "step": 3705
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.139359698681733e-05,
+      "loss": 0.2847,
+      "step": 3706
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1391242937853107e-05,
+      "loss": 0.3842,
+      "step": 3707
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.138888888888889e-05,
+      "loss": 0.2776,
+      "step": 3708
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1386534839924674e-05,
+      "loss": 0.3706,
+      "step": 3709
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.138418079096045e-05,
+      "loss": 0.3357,
+      "step": 3710
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1381826741996236e-05,
+      "loss": 0.3832,
+      "step": 3711
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.137947269303202e-05,
+      "loss": 0.3592,
+      "step": 3712
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1377118644067797e-05,
+      "loss": 0.2515,
+      "step": 3713
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.137476459510358e-05,
+      "loss": 0.3177,
+      "step": 3714
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1372410546139364e-05,
+      "loss": 0.3099,
+      "step": 3715
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.137005649717514e-05,
+      "loss": 0.3384,
+      "step": 3716
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1367702448210926e-05,
+      "loss": 0.4076,
+      "step": 3717
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.136534839924671e-05,
+      "loss": 0.2519,
+      "step": 3718
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1362994350282487e-05,
+      "loss": 0.319,
+      "step": 3719
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.136064030131827e-05,
+      "loss": 0.2549,
+      "step": 3720
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.135828625235405e-05,
+      "loss": 0.3299,
+      "step": 3721
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.135593220338983e-05,
+      "loss": 0.3057,
+      "step": 3722
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1353578154425616e-05,
+      "loss": 0.3441,
+      "step": 3723
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.135122410546139e-05,
+      "loss": 0.2855,
+      "step": 3724
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1348870056497177e-05,
+      "loss": 0.3675,
+      "step": 3725
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.134651600753296e-05,
+      "loss": 0.3261,
+      "step": 3726
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.134416195856874e-05,
+      "loss": 0.4974,
+      "step": 3727
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.134180790960452e-05,
+      "loss": 0.3834,
+      "step": 3728
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1339453860640306e-05,
+      "loss": 0.3359,
+      "step": 3729
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.133709981167608e-05,
+      "loss": 0.379,
+      "step": 3730
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1334745762711867e-05,
+      "loss": 0.3222,
+      "step": 3731
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.133239171374765e-05,
+      "loss": 0.302,
+      "step": 3732
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.133003766478343e-05,
+      "loss": 0.2918,
+      "step": 3733
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.132768361581921e-05,
+      "loss": 0.3116,
+      "step": 3734
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1325329566854996e-05,
+      "loss": 0.3656,
+      "step": 3735
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.132297551789077e-05,
+      "loss": 0.4063,
+      "step": 3736
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.1320621468926557e-05,
+      "loss": 0.182,
+      "step": 3737
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.1318267419962334e-05,
+      "loss": 0.2696,
+      "step": 3738
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.131591337099812e-05,
+      "loss": 0.4208,
+      "step": 3739
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.13135593220339e-05,
+      "loss": 0.3508,
+      "step": 3740
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.131120527306968e-05,
+      "loss": 0.3113,
+      "step": 3741
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.130885122410546e-05,
+      "loss": 0.4219,
+      "step": 3742
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.1306497175141247e-05,
+      "loss": 0.3859,
+      "step": 3743
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.1304143126177024e-05,
+      "loss": 0.2985,
+      "step": 3744
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.130178907721281e-05,
+      "loss": 0.4313,
+      "step": 3745
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.129943502824859e-05,
+      "loss": 0.4577,
+      "step": 3746
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.129708097928437e-05,
+      "loss": 0.4215,
+      "step": 3747
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.129472693032015e-05,
+      "loss": 0.2643,
+      "step": 3748
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.1292372881355937e-05,
+      "loss": 0.2658,
+      "step": 3749
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.1290018832391714e-05,
+      "loss": 0.406,
+      "step": 3750
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.12876647834275e-05,
+      "loss": 0.3976,
+      "step": 3751
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.128531073446328e-05,
+      "loss": 0.3477,
+      "step": 3752
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.128295668549906e-05,
+      "loss": 0.2937,
+      "step": 3753
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.128060263653484e-05,
+      "loss": 0.3756,
+      "step": 3754
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.127824858757062e-05,
+      "loss": 0.3364,
+      "step": 3755
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.127589453860641e-05,
+      "loss": 0.4324,
+      "step": 3756
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.127354048964219e-05,
+      "loss": 0.3508,
+      "step": 3757
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1271186440677965e-05,
+      "loss": 0.4147,
+      "step": 3758
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.126883239171375e-05,
+      "loss": 0.3941,
+      "step": 3759
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.126647834274953e-05,
+      "loss": 0.2753,
+      "step": 3760
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.126412429378531e-05,
+      "loss": 0.3193,
+      "step": 3761
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1261770244821094e-05,
+      "loss": 0.293,
+      "step": 3762
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.125941619585688e-05,
+      "loss": 0.3931,
+      "step": 3763
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1257062146892655e-05,
+      "loss": 0.3706,
+      "step": 3764
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.125470809792844e-05,
+      "loss": 0.2613,
+      "step": 3765
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.125235404896422e-05,
+      "loss": 0.3778,
+      "step": 3766
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.125e-05,
+      "loss": 0.2308,
+      "step": 3767
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1247645951035784e-05,
+      "loss": 0.4068,
+      "step": 3768
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.124529190207157e-05,
+      "loss": 0.3936,
+      "step": 3769
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1242937853107345e-05,
+      "loss": 0.3691,
+      "step": 3770
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.124058380414313e-05,
+      "loss": 0.3389,
+      "step": 3771
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1238229755178906e-05,
+      "loss": 0.2693,
+      "step": 3772
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1235875706214697e-05,
+      "loss": 0.2658,
+      "step": 3773
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1233521657250474e-05,
+      "loss": 0.3653,
+      "step": 3774
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.123116760828625e-05,
+      "loss": 0.3527,
+      "step": 3775
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1228813559322035e-05,
+      "loss": 0.3895,
+      "step": 3776
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.122645951035782e-05,
+      "loss": 0.275,
+      "step": 3777
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1224105461393596e-05,
+      "loss": 0.415,
+      "step": 3778
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.122175141242938e-05,
+      "loss": 0.3604,
+      "step": 3779
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1219397363465164e-05,
+      "loss": 0.2926,
+      "step": 3780
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.121704331450094e-05,
+      "loss": 0.2788,
+      "step": 3781
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1214689265536725e-05,
+      "loss": 0.2117,
+      "step": 3782
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.121233521657251e-05,
+      "loss": 0.2379,
+      "step": 3783
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1209981167608286e-05,
+      "loss": 0.3198,
+      "step": 3784
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.120762711864407e-05,
+      "loss": 0.2637,
+      "step": 3785
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1205273069679854e-05,
+      "loss": 0.3162,
+      "step": 3786
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.120291902071563e-05,
+      "loss": 0.4621,
+      "step": 3787
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1200564971751415e-05,
+      "loss": 0.2862,
+      "step": 3788
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.119821092278719e-05,
+      "loss": 0.3977,
+      "step": 3789
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.119585687382298e-05,
+      "loss": 0.3807,
+      "step": 3790
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.119350282485876e-05,
+      "loss": 0.3578,
+      "step": 3791
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.119114877589454e-05,
+      "loss": 0.169,
+      "step": 3792
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.118879472693033e-05,
+      "loss": 0.3856,
+      "step": 3793
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1186440677966105e-05,
+      "loss": 0.3446,
+      "step": 3794
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.118408662900188e-05,
+      "loss": 0.3474,
+      "step": 3795
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1181732580037666e-05,
+      "loss": 0.2823,
+      "step": 3796
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.117937853107345e-05,
+      "loss": 0.2553,
+      "step": 3797
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.117702448210923e-05,
+      "loss": 0.2717,
+      "step": 3798
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.117467043314501e-05,
+      "loss": 0.286,
+      "step": 3799
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1172316384180795e-05,
+      "loss": 0.2908,
+      "step": 3800
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.116996233521657e-05,
+      "loss": 0.4076,
+      "step": 3801
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1167608286252356e-05,
+      "loss": 0.3347,
+      "step": 3802
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.116525423728814e-05,
+      "loss": 0.3313,
+      "step": 3803
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1162900188323924e-05,
+      "loss": 0.2896,
+      "step": 3804
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.11605461393597e-05,
+      "loss": 0.3105,
+      "step": 3805
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.115819209039548e-05,
+      "loss": 0.381,
+      "step": 3806
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.115583804143127e-05,
+      "loss": 0.3326,
+      "step": 3807
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1153483992467046e-05,
+      "loss": 0.2198,
+      "step": 3808
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.115112994350282e-05,
+      "loss": 0.2879,
+      "step": 3809
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1148775894538614e-05,
+      "loss": 0.443,
+      "step": 3810
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.114642184557439e-05,
+      "loss": 0.3747,
+      "step": 3811
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.114406779661017e-05,
+      "loss": 0.4125,
+      "step": 3812
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.114171374764595e-05,
+      "loss": 0.438,
+      "step": 3813
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1139359698681736e-05,
+      "loss": 0.2555,
+      "step": 3814
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.113700564971751e-05,
+      "loss": 0.4151,
+      "step": 3815
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.11346516007533e-05,
+      "loss": 0.3101,
+      "step": 3816
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.113229755178908e-05,
+      "loss": 0.3484,
+      "step": 3817
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.112994350282486e-05,
+      "loss": 0.1995,
+      "step": 3818
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.112758945386064e-05,
+      "loss": 0.2558,
+      "step": 3819
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1125235404896426e-05,
+      "loss": 0.225,
+      "step": 3820
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.112288135593221e-05,
+      "loss": 0.2593,
+      "step": 3821
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.112052730696799e-05,
+      "loss": 0.2417,
+      "step": 3822
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.1118173258003764e-05,
+      "loss": 0.4145,
+      "step": 3823
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.1115819209039555e-05,
+      "loss": 0.3914,
+      "step": 3824
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.111346516007533e-05,
+      "loss": 0.2144,
+      "step": 3825
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.111111111111111e-05,
+      "loss": 0.2748,
+      "step": 3826
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.11087570621469e-05,
+      "loss": 0.3054,
+      "step": 3827
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.110640301318268e-05,
+      "loss": 0.3674,
+      "step": 3828
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.1104048964218454e-05,
+      "loss": 0.3268,
+      "step": 3829
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.110169491525424e-05,
+      "loss": 0.3073,
+      "step": 3830
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.109934086629002e-05,
+      "loss": 0.3917,
+      "step": 3831
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.10969868173258e-05,
+      "loss": 0.3872,
+      "step": 3832
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.109463276836158e-05,
+      "loss": 0.2811,
+      "step": 3833
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.109227871939737e-05,
+      "loss": 0.2092,
+      "step": 3834
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.108992467043315e-05,
+      "loss": 0.3113,
+      "step": 3835
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.108757062146893e-05,
+      "loss": 0.2555,
+      "step": 3836
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.108521657250471e-05,
+      "loss": 0.4124,
+      "step": 3837
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.1082862523540496e-05,
+      "loss": 0.452,
+      "step": 3838
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.108050847457627e-05,
+      "loss": 0.3926,
+      "step": 3839
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.107815442561205e-05,
+      "loss": 0.324,
+      "step": 3840
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.107580037664784e-05,
+      "loss": 0.2636,
+      "step": 3841
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.107344632768362e-05,
+      "loss": 0.375,
+      "step": 3842
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.1071092278719395e-05,
+      "loss": 0.4922,
+      "step": 3843
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.1068738229755186e-05,
+      "loss": 0.2856,
+      "step": 3844
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.106638418079096e-05,
+      "loss": 0.4623,
+      "step": 3845
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.106403013182674e-05,
+      "loss": 0.4265,
+      "step": 3846
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.1061676082862524e-05,
+      "loss": 0.41,
+      "step": 3847
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.105932203389831e-05,
+      "loss": 0.3249,
+      "step": 3848
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.1056967984934085e-05,
+      "loss": 0.4372,
+      "step": 3849
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.105461393596987e-05,
+      "loss": 0.193,
+      "step": 3850
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.105225988700565e-05,
+      "loss": 0.2949,
+      "step": 3851
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.104990583804144e-05,
+      "loss": 0.3255,
+      "step": 3852
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.1047551789077214e-05,
+      "loss": 0.4105,
+      "step": 3853
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.1045197740113e-05,
+      "loss": 0.3291,
+      "step": 3854
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.104284369114878e-05,
+      "loss": 0.2861,
+      "step": 3855
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.104048964218456e-05,
+      "loss": 0.3837,
+      "step": 3856
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.1038135593220336e-05,
+      "loss": 0.4354,
+      "step": 3857
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.103578154425613e-05,
+      "loss": 0.2645,
+      "step": 3858
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.1033427495291904e-05,
+      "loss": 0.241,
+      "step": 3859
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.103107344632768e-05,
+      "loss": 0.4367,
+      "step": 3860
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.102871939736347e-05,
+      "loss": 0.2625,
+      "step": 3861
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.102636534839925e-05,
+      "loss": 0.3016,
+      "step": 3862
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.1024011299435026e-05,
+      "loss": 0.3225,
+      "step": 3863
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.102165725047081e-05,
+      "loss": 0.3062,
+      "step": 3864
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1019303201506594e-05,
+      "loss": 0.3372,
+      "step": 3865
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.101694915254237e-05,
+      "loss": 0.3093,
+      "step": 3866
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1014595103578155e-05,
+      "loss": 0.2754,
+      "step": 3867
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.101224105461394e-05,
+      "loss": 0.3875,
+      "step": 3868
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.100988700564972e-05,
+      "loss": 0.3933,
+      "step": 3869
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.10075329566855e-05,
+      "loss": 0.2748,
+      "step": 3870
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1005178907721284e-05,
+      "loss": 0.3746,
+      "step": 3871
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.100282485875707e-05,
+      "loss": 0.3024,
+      "step": 3872
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1000470809792845e-05,
+      "loss": 0.3646,
+      "step": 3873
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.099811676082862e-05,
+      "loss": 0.2623,
+      "step": 3874
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.099576271186441e-05,
+      "loss": 0.3259,
+      "step": 3875
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.099340866290019e-05,
+      "loss": 0.2963,
+      "step": 3876
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.099105461393597e-05,
+      "loss": 0.3251,
+      "step": 3877
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.098870056497176e-05,
+      "loss": 0.326,
+      "step": 3878
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0986346516007535e-05,
+      "loss": 0.3634,
+      "step": 3879
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.098399246704331e-05,
+      "loss": 0.2534,
+      "step": 3880
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0981638418079096e-05,
+      "loss": 0.4424,
+      "step": 3881
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.097928436911488e-05,
+      "loss": 0.3983,
+      "step": 3882
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0976930320150664e-05,
+      "loss": 0.3034,
+      "step": 3883
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.097457627118644e-05,
+      "loss": 0.2319,
+      "step": 3884
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0972222222222225e-05,
+      "loss": 0.4273,
+      "step": 3885
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.096986817325801e-05,
+      "loss": 0.4201,
+      "step": 3886
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.0967514124293786e-05,
+      "loss": 0.3083,
+      "step": 3887
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.096516007532957e-05,
+      "loss": 0.2564,
+      "step": 3888
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.0962806026365354e-05,
+      "loss": 0.2283,
+      "step": 3889
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.096045197740113e-05,
+      "loss": 0.3893,
+      "step": 3890
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.095809792843691e-05,
+      "loss": 0.2499,
+      "step": 3891
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.09557438794727e-05,
+      "loss": 0.4107,
+      "step": 3892
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.0953389830508476e-05,
+      "loss": 0.4139,
+      "step": 3893
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.095103578154425e-05,
+      "loss": 0.2816,
+      "step": 3894
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.0948681732580044e-05,
+      "loss": 0.3475,
+      "step": 3895
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.094632768361582e-05,
+      "loss": 0.2481,
+      "step": 3896
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.09439736346516e-05,
+      "loss": 0.4145,
+      "step": 3897
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.094161958568738e-05,
+      "loss": 0.2912,
+      "step": 3898
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.0939265536723166e-05,
+      "loss": 0.3049,
+      "step": 3899
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.093691148775895e-05,
+      "loss": 0.2344,
+      "step": 3900
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.093455743879473e-05,
+      "loss": 0.3539,
+      "step": 3901
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.093220338983051e-05,
+      "loss": 0.3548,
+      "step": 3902
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.0929849340866295e-05,
+      "loss": 0.3616,
+      "step": 3903
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.092749529190207e-05,
+      "loss": 0.3833,
+      "step": 3904
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.0925141242937856e-05,
+      "loss": 0.5541,
+      "step": 3905
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.092278719397364e-05,
+      "loss": 0.37,
+      "step": 3906
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.092043314500942e-05,
+      "loss": 0.4103,
+      "step": 3907
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0918079096045194e-05,
+      "loss": 0.2338,
+      "step": 3908
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0915725047080985e-05,
+      "loss": 0.3992,
+      "step": 3909
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.091337099811676e-05,
+      "loss": 0.3082,
+      "step": 3910
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.091101694915254e-05,
+      "loss": 0.3155,
+      "step": 3911
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.090866290018833e-05,
+      "loss": 0.2585,
+      "step": 3912
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.090630885122411e-05,
+      "loss": 0.2913,
+      "step": 3913
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.090395480225989e-05,
+      "loss": 0.4282,
+      "step": 3914
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.090160075329567e-05,
+      "loss": 0.3979,
+      "step": 3915
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.089924670433145e-05,
+      "loss": 0.4054,
+      "step": 3916
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0896892655367236e-05,
+      "loss": 0.3468,
+      "step": 3917
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.089453860640301e-05,
+      "loss": 0.2972,
+      "step": 3918
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.08921845574388e-05,
+      "loss": 0.2338,
+      "step": 3919
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.088983050847458e-05,
+      "loss": 0.2222,
+      "step": 3920
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.088747645951036e-05,
+      "loss": 0.3032,
+      "step": 3921
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.088512241054614e-05,
+      "loss": 0.3055,
+      "step": 3922
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0882768361581926e-05,
+      "loss": 0.3839,
+      "step": 3923
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.08804143126177e-05,
+      "loss": 0.3487,
+      "step": 3924
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.087806026365348e-05,
+      "loss": 0.2537,
+      "step": 3925
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.087570621468927e-05,
+      "loss": 0.4674,
+      "step": 3926
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.087335216572505e-05,
+      "loss": 0.3391,
+      "step": 3927
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0870998116760825e-05,
+      "loss": 0.2722,
+      "step": 3928
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.0868644067796616e-05,
+      "loss": 0.3276,
+      "step": 3929
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.086629001883239e-05,
+      "loss": 0.4798,
+      "step": 3930
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.086393596986818e-05,
+      "loss": 0.2786,
+      "step": 3931
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.0861581920903954e-05,
+      "loss": 0.2328,
+      "step": 3932
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.085922787193974e-05,
+      "loss": 0.2901,
+      "step": 3933
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.085687382297552e-05,
+      "loss": 0.4259,
+      "step": 3934
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.08545197740113e-05,
+      "loss": 0.3105,
+      "step": 3935
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.085216572504708e-05,
+      "loss": 0.4319,
+      "step": 3936
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.084981167608287e-05,
+      "loss": 0.3159,
+      "step": 3937
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.0847457627118644e-05,
+      "loss": 0.4941,
+      "step": 3938
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.084510357815443e-05,
+      "loss": 0.3292,
+      "step": 3939
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.084274952919021e-05,
+      "loss": 0.3927,
+      "step": 3940
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.084039548022599e-05,
+      "loss": 0.3256,
+      "step": 3941
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.083804143126177e-05,
+      "loss": 0.3369,
+      "step": 3942
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.083568738229756e-05,
+      "loss": 0.3514,
+      "step": 3943
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.0833333333333334e-05,
+      "loss": 0.219,
+      "step": 3944
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.083097928436911e-05,
+      "loss": 0.417,
+      "step": 3945
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.08286252354049e-05,
+      "loss": 0.4043,
+      "step": 3946
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.082627118644068e-05,
+      "loss": 0.3731,
+      "step": 3947
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.082391713747646e-05,
+      "loss": 0.2747,
+      "step": 3948
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.082156308851224e-05,
+      "loss": 0.3463,
+      "step": 3949
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0819209039548024e-05,
+      "loss": 0.3274,
+      "step": 3950
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.081685499058381e-05,
+      "loss": 0.327,
+      "step": 3951
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0814500941619585e-05,
+      "loss": 0.3824,
+      "step": 3952
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.081214689265537e-05,
+      "loss": 0.3412,
+      "step": 3953
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.080979284369115e-05,
+      "loss": 0.3981,
+      "step": 3954
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.080743879472693e-05,
+      "loss": 0.2758,
+      "step": 3955
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0805084745762714e-05,
+      "loss": 0.2184,
+      "step": 3956
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.08027306967985e-05,
+      "loss": 0.4266,
+      "step": 3957
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0800376647834275e-05,
+      "loss": 0.4013,
+      "step": 3958
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.079802259887006e-05,
+      "loss": 0.3451,
+      "step": 3959
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.079566854990584e-05,
+      "loss": 0.2536,
+      "step": 3960
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.079331450094162e-05,
+      "loss": 0.3675,
+      "step": 3961
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0790960451977404e-05,
+      "loss": 0.4952,
+      "step": 3962
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.078860640301319e-05,
+      "loss": 0.3285,
+      "step": 3963
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0786252354048965e-05,
+      "loss": 0.4946,
+      "step": 3964
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.078389830508475e-05,
+      "loss": 0.3073,
+      "step": 3965
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0781544256120526e-05,
+      "loss": 0.2401,
+      "step": 3966
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.077919020715631e-05,
+      "loss": 0.1595,
+      "step": 3967
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0776836158192094e-05,
+      "loss": 0.262,
+      "step": 3968
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.077448210922787e-05,
+      "loss": 0.2517,
+      "step": 3969
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0772128060263655e-05,
+      "loss": 0.3074,
+      "step": 3970
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.076977401129944e-05,
+      "loss": 0.3043,
+      "step": 3971
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0767419962335216e-05,
+      "loss": 0.4418,
+      "step": 3972
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0765065913371e-05,
+      "loss": 0.3923,
+      "step": 3973
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0762711864406784e-05,
+      "loss": 0.4242,
+      "step": 3974
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.076035781544256e-05,
+      "loss": 0.3371,
+      "step": 3975
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0758003766478345e-05,
+      "loss": 0.2932,
+      "step": 3976
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.075564971751413e-05,
+      "loss": 0.3478,
+      "step": 3977
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0753295668549906e-05,
+      "loss": 0.4157,
+      "step": 3978
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.075094161958569e-05,
+      "loss": 0.4641,
+      "step": 3979
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0748587570621474e-05,
+      "loss": 0.2319,
+      "step": 3980
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.074623352165725e-05,
+      "loss": 0.259,
+      "step": 3981
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0743879472693035e-05,
+      "loss": 0.239,
+      "step": 3982
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.074152542372881e-05,
+      "loss": 0.3564,
+      "step": 3983
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0739171374764596e-05,
+      "loss": 0.2858,
+      "step": 3984
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.073681732580038e-05,
+      "loss": 0.2963,
+      "step": 3985
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.073446327683616e-05,
+      "loss": 0.3234,
+      "step": 3986
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.073210922787194e-05,
+      "loss": 0.3731,
+      "step": 3987
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0729755178907725e-05,
+      "loss": 0.2214,
+      "step": 3988
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.07274011299435e-05,
+      "loss": 0.2631,
+      "step": 3989
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0725047080979286e-05,
+      "loss": 0.3288,
+      "step": 3990
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.072269303201507e-05,
+      "loss": 0.33,
+      "step": 3991
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.072033898305085e-05,
+      "loss": 0.302,
+      "step": 3992
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.071798493408663e-05,
+      "loss": 0.2778,
+      "step": 3993
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0715630885122415e-05,
+      "loss": 0.2978,
+      "step": 3994
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.071327683615819e-05,
+      "loss": 0.2444,
+      "step": 3995
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0710922787193976e-05,
+      "loss": 0.4302,
+      "step": 3996
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.070856873822976e-05,
+      "loss": 0.3779,
+      "step": 3997
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.070621468926554e-05,
+      "loss": 0.4369,
+      "step": 3998
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.070386064030132e-05,
+      "loss": 0.5371,
+      "step": 3999
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.07015065913371e-05,
+      "loss": 0.3552,
+      "step": 4000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.069915254237288e-05,
+      "loss": 0.3527,
+      "step": 4001
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0696798493408666e-05,
+      "loss": 0.2474,
+      "step": 4002
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0694444444444444e-05,
+      "loss": 0.4436,
+      "step": 4003
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.069209039548023e-05,
+      "loss": 0.4707,
+      "step": 4004
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.068973634651601e-05,
+      "loss": 0.3694,
+      "step": 4005
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.068738229755179e-05,
+      "loss": 0.4218,
+      "step": 4006
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.068502824858757e-05,
+      "loss": 0.2799,
+      "step": 4007
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0682674199623356e-05,
+      "loss": 0.2431,
+      "step": 4008
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0680320150659134e-05,
+      "loss": 0.4536,
+      "step": 4009
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.067796610169492e-05,
+      "loss": 0.3038,
+      "step": 4010
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.06756120527307e-05,
+      "loss": 0.4319,
+      "step": 4011
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.067325800376648e-05,
+      "loss": 0.3389,
+      "step": 4012
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.067090395480226e-05,
+      "loss": 0.5023,
+      "step": 4013
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0668549905838046e-05,
+      "loss": 0.333,
+      "step": 4014
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0666195856873824e-05,
+      "loss": 0.2637,
+      "step": 4015
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.066384180790961e-05,
+      "loss": 0.4157,
+      "step": 4016
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.066148775894539e-05,
+      "loss": 0.3155,
+      "step": 4017
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.065913370998117e-05,
+      "loss": 0.2457,
+      "step": 4018
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.065677966101695e-05,
+      "loss": 0.3484,
+      "step": 4019
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.065442561205273e-05,
+      "loss": 0.3363,
+      "step": 4020
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0652071563088513e-05,
+      "loss": 0.3313,
+      "step": 4021
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.06497175141243e-05,
+      "loss": 0.3192,
+      "step": 4022
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0647363465160075e-05,
+      "loss": 0.2936,
+      "step": 4023
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.064500941619586e-05,
+      "loss": 0.4256,
+      "step": 4024
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.064265536723164e-05,
+      "loss": 0.4565,
+      "step": 4025
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.064030131826742e-05,
+      "loss": 0.3231,
+      "step": 4026
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0637947269303203e-05,
+      "loss": 0.3051,
+      "step": 4027
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.063559322033899e-05,
+      "loss": 0.4167,
+      "step": 4028
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0633239171374765e-05,
+      "loss": 0.3367,
+      "step": 4029
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.063088512241055e-05,
+      "loss": 0.3292,
+      "step": 4030
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.062853107344633e-05,
+      "loss": 0.2795,
+      "step": 4031
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.062617702448211e-05,
+      "loss": 0.3199,
+      "step": 4032
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0623822975517893e-05,
+      "loss": 0.3875,
+      "step": 4033
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.062146892655368e-05,
+      "loss": 0.1801,
+      "step": 4034
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0619114877589455e-05,
+      "loss": 0.3432,
+      "step": 4035
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.061676082862524e-05,
+      "loss": 0.4207,
+      "step": 4036
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0614406779661016e-05,
+      "loss": 0.3677,
+      "step": 4037
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.06120527306968e-05,
+      "loss": 0.2848,
+      "step": 4038
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0609698681732583e-05,
+      "loss": 0.5048,
+      "step": 4039
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.060734463276836e-05,
+      "loss": 0.3228,
+      "step": 4040
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0604990583804145e-05,
+      "loss": 0.4409,
+      "step": 4041
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.060263653483993e-05,
+      "loss": 0.3238,
+      "step": 4042
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0600282485875706e-05,
+      "loss": 0.2269,
+      "step": 4043
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.059792843691149e-05,
+      "loss": 0.3655,
+      "step": 4044
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0595574387947273e-05,
+      "loss": 0.3629,
+      "step": 4045
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.059322033898305e-05,
+      "loss": 0.373,
+      "step": 4046
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0590866290018835e-05,
+      "loss": 0.2742,
+      "step": 4047
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.058851224105462e-05,
+      "loss": 0.246,
+      "step": 4048
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0586158192090396e-05,
+      "loss": 0.3108,
+      "step": 4049
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.058380414312618e-05,
+      "loss": 0.2551,
+      "step": 4050
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0581450094161963e-05,
+      "loss": 0.3146,
+      "step": 4051
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.057909604519774e-05,
+      "loss": 0.2494,
+      "step": 4052
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0576741996233525e-05,
+      "loss": 0.3607,
+      "step": 4053
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.05743879472693e-05,
+      "loss": 0.3487,
+      "step": 4054
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0572033898305086e-05,
+      "loss": 0.2877,
+      "step": 4055
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.056967984934087e-05,
+      "loss": 0.3115,
+      "step": 4056
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.056732580037665e-05,
+      "loss": 0.4524,
+      "step": 4057
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.056497175141243e-05,
+      "loss": 0.2934,
+      "step": 4058
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0562617702448215e-05,
+      "loss": 0.243,
+      "step": 4059
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.056026365348399e-05,
+      "loss": 0.3829,
+      "step": 4060
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0557909604519776e-05,
+      "loss": 0.2745,
+      "step": 4061
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.055555555555556e-05,
+      "loss": 0.449,
+      "step": 4062
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.055320150659134e-05,
+      "loss": 0.3634,
+      "step": 4063
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.055084745762712e-05,
+      "loss": 0.3846,
+      "step": 4064
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0548493408662905e-05,
+      "loss": 0.3212,
+      "step": 4065
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.054613935969868e-05,
+      "loss": 0.3974,
+      "step": 4066
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0543785310734466e-05,
+      "loss": 0.2839,
+      "step": 4067
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.054143126177025e-05,
+      "loss": 0.2755,
+      "step": 4068
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.053907721280603e-05,
+      "loss": 0.4286,
+      "step": 4069
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.053672316384181e-05,
+      "loss": 0.3433,
+      "step": 4070
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.053436911487759e-05,
+      "loss": 0.2627,
+      "step": 4071
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.053201506591337e-05,
+      "loss": 0.3768,
+      "step": 4072
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0529661016949156e-05,
+      "loss": 0.4877,
+      "step": 4073
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.052730696798493e-05,
+      "loss": 0.2945,
+      "step": 4074
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.052495291902072e-05,
+      "loss": 0.3174,
+      "step": 4075
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.05225988700565e-05,
+      "loss": 0.3846,
+      "step": 4076
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.052024482109228e-05,
+      "loss": 0.3584,
+      "step": 4077
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.051789077212806e-05,
+      "loss": 0.2786,
+      "step": 4078
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.0515536723163846e-05,
+      "loss": 0.3202,
+      "step": 4079
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.051318267419962e-05,
+      "loss": 0.3457,
+      "step": 4080
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.051082862523541e-05,
+      "loss": 0.3405,
+      "step": 4081
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.050847457627119e-05,
+      "loss": 0.3477,
+      "step": 4082
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.050612052730697e-05,
+      "loss": 0.2889,
+      "step": 4083
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.050376647834275e-05,
+      "loss": 0.3,
+      "step": 4084
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.0501412429378536e-05,
+      "loss": 0.182,
+      "step": 4085
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.049905838041431e-05,
+      "loss": 0.3993,
+      "step": 4086
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.04967043314501e-05,
+      "loss": 0.4038,
+      "step": 4087
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.0494350282485874e-05,
+      "loss": 0.485,
+      "step": 4088
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.0491996233521665e-05,
+      "loss": 0.2953,
+      "step": 4089
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.048964218455744e-05,
+      "loss": 0.3622,
+      "step": 4090
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.048728813559322e-05,
+      "loss": 0.461,
+      "step": 4091
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.0484934086629e-05,
+      "loss": 0.3463,
+      "step": 4092
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.048258003766479e-05,
+      "loss": 0.2809,
+      "step": 4093
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.0480225988700564e-05,
+      "loss": 0.2423,
+      "step": 4094
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.047787193973635e-05,
+      "loss": 0.3256,
+      "step": 4095
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.047551789077213e-05,
+      "loss": 0.423,
+      "step": 4096
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.047316384180791e-05,
+      "loss": 0.2598,
+      "step": 4097
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.047080979284369e-05,
+      "loss": 0.4071,
+      "step": 4098
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.046845574387948e-05,
+      "loss": 0.3208,
+      "step": 4099
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0466101694915254e-05,
+      "loss": 0.3755,
+      "step": 4100
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.046374764595104e-05,
+      "loss": 0.3786,
+      "step": 4101
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.046139359698682e-05,
+      "loss": 0.3978,
+      "step": 4102
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.04590395480226e-05,
+      "loss": 0.4359,
+      "step": 4103
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.045668549905838e-05,
+      "loss": 0.3159,
+      "step": 4104
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.045433145009416e-05,
+      "loss": 0.2923,
+      "step": 4105
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.045197740112995e-05,
+      "loss": 0.3218,
+      "step": 4106
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.044962335216573e-05,
+      "loss": 0.3317,
+      "step": 4107
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0447269303201505e-05,
+      "loss": 0.396,
+      "step": 4108
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0444915254237296e-05,
+      "loss": 0.3119,
+      "step": 4109
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.044256120527307e-05,
+      "loss": 0.2331,
+      "step": 4110
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.044020715630885e-05,
+      "loss": 0.3991,
+      "step": 4111
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0437853107344634e-05,
+      "loss": 0.3683,
+      "step": 4112
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.043549905838042e-05,
+      "loss": 0.2445,
+      "step": 4113
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0433145009416195e-05,
+      "loss": 0.3701,
+      "step": 4114
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.043079096045198e-05,
+      "loss": 0.2558,
+      "step": 4115
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.042843691148776e-05,
+      "loss": 0.2462,
+      "step": 4116
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.042608286252354e-05,
+      "loss": 0.2542,
+      "step": 4117
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0423728813559324e-05,
+      "loss": 0.2379,
+      "step": 4118
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.042137476459511e-05,
+      "loss": 0.343,
+      "step": 4119
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.041902071563089e-05,
+      "loss": 0.2379,
+      "step": 4120
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.041666666666667e-05,
+      "loss": 0.3339,
+      "step": 4121
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0414312617702446e-05,
+      "loss": 0.2884,
+      "step": 4122
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.041195856873824e-05,
+      "loss": 0.2946,
+      "step": 4123
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0409604519774014e-05,
+      "loss": 0.3045,
+      "step": 4124
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.040725047080979e-05,
+      "loss": 0.3311,
+      "step": 4125
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.040489642184558e-05,
+      "loss": 0.2595,
+      "step": 4126
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.040254237288136e-05,
+      "loss": 0.3373,
+      "step": 4127
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0400188323917136e-05,
+      "loss": 0.2607,
+      "step": 4128
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.039783427495292e-05,
+      "loss": 0.2906,
+      "step": 4129
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0395480225988704e-05,
+      "loss": 0.3553,
+      "step": 4130
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.039312617702448e-05,
+      "loss": 0.3675,
+      "step": 4131
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0390772128060265e-05,
+      "loss": 0.3345,
+      "step": 4132
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.038841807909605e-05,
+      "loss": 0.3459,
+      "step": 4133
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0386064030131826e-05,
+      "loss": 0.3159,
+      "step": 4134
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.038370998116761e-05,
+      "loss": 0.3895,
+      "step": 4135
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0381355932203394e-05,
+      "loss": 0.2382,
+      "step": 4136
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.037900188323918e-05,
+      "loss": 0.2767,
+      "step": 4137
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0376647834274955e-05,
+      "loss": 0.2279,
+      "step": 4138
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.037429378531073e-05,
+      "loss": 0.2259,
+      "step": 4139
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.037193973634652e-05,
+      "loss": 0.4026,
+      "step": 4140
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.03695856873823e-05,
+      "loss": 0.367,
+      "step": 4141
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.036723163841808e-05,
+      "loss": 0.3445,
+      "step": 4142
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.036487758945387e-05,
+      "loss": 0.2946,
+      "step": 4143
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0362523540489645e-05,
+      "loss": 0.2931,
+      "step": 4144
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.036016949152542e-05,
+      "loss": 0.41,
+      "step": 4145
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0357815442561206e-05,
+      "loss": 0.4136,
+      "step": 4146
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.035546139359699e-05,
+      "loss": 0.3578,
+      "step": 4147
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.035310734463277e-05,
+      "loss": 0.4197,
+      "step": 4148
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.035075329566855e-05,
+      "loss": 0.3107,
+      "step": 4149
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0348399246704335e-05,
+      "loss": 0.3232,
+      "step": 4150
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.034604519774011e-05,
+      "loss": 0.4379,
+      "step": 4151
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0343691148775896e-05,
+      "loss": 0.3631,
+      "step": 4152
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.034133709981168e-05,
+      "loss": 0.2728,
+      "step": 4153
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0338983050847464e-05,
+      "loss": 0.3852,
+      "step": 4154
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.033662900188324e-05,
+      "loss": 0.3041,
+      "step": 4155
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.033427495291902e-05,
+      "loss": 0.4163,
+      "step": 4156
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.033192090395481e-05,
+      "loss": 0.2762,
+      "step": 4157
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0329566854990586e-05,
+      "loss": 0.1958,
+      "step": 4158
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.032721280602636e-05,
+      "loss": 0.316,
+      "step": 4159
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0324858757062154e-05,
+      "loss": 0.2706,
+      "step": 4160
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.032250470809793e-05,
+      "loss": 0.3394,
+      "step": 4161
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.032015065913371e-05,
+      "loss": 0.2022,
+      "step": 4162
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.031779661016949e-05,
+      "loss": 0.2266,
+      "step": 4163
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0315442561205276e-05,
+      "loss": 0.2958,
+      "step": 4164
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.031308851224105e-05,
+      "loss": 0.5119,
+      "step": 4165
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.031073446327684e-05,
+      "loss": 0.2262,
+      "step": 4166
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.030838041431262e-05,
+      "loss": 0.1969,
+      "step": 4167
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0306026365348405e-05,
+      "loss": 0.2742,
+      "step": 4168
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.030367231638418e-05,
+      "loss": 0.3414,
+      "step": 4169
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0301318267419966e-05,
+      "loss": 0.3796,
+      "step": 4170
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.029896421845575e-05,
+      "loss": 0.2796,
+      "step": 4171
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.029661016949153e-05,
+      "loss": 0.4311,
+      "step": 4172
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0294256120527304e-05,
+      "loss": 0.3744,
+      "step": 4173
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0291902071563095e-05,
+      "loss": 0.4297,
+      "step": 4174
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.028954802259887e-05,
+      "loss": 0.3757,
+      "step": 4175
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.028719397363465e-05,
+      "loss": 0.2206,
+      "step": 4176
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.028483992467044e-05,
+      "loss": 0.349,
+      "step": 4177
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.028248587570622e-05,
+      "loss": 0.1441,
+      "step": 4178
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0280131826741994e-05,
+      "loss": 0.3211,
+      "step": 4179
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.027777777777778e-05,
+      "loss": 0.2869,
+      "step": 4180
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.027542372881356e-05,
+      "loss": 0.2487,
+      "step": 4181
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.027306967984934e-05,
+      "loss": 0.273,
+      "step": 4182
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.027071563088512e-05,
+      "loss": 0.241,
+      "step": 4183
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.026836158192091e-05,
+      "loss": 0.2336,
+      "step": 4184
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.026600753295669e-05,
+      "loss": 0.2987,
+      "step": 4185
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.026365348399247e-05,
+      "loss": 0.3111,
+      "step": 4186
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.026129943502825e-05,
+      "loss": 0.2296,
+      "step": 4187
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0258945386064036e-05,
+      "loss": 0.292,
+      "step": 4188
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.025659133709981e-05,
+      "loss": 0.3515,
+      "step": 4189
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.025423728813559e-05,
+      "loss": 0.4625,
+      "step": 4190
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.025188323917138e-05,
+      "loss": 0.2747,
+      "step": 4191
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.024952919020716e-05,
+      "loss": 0.4395,
+      "step": 4192
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0247175141242935e-05,
+      "loss": 0.3793,
+      "step": 4193
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0244821092278726e-05,
+      "loss": 0.4036,
+      "step": 4194
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.02424670433145e-05,
+      "loss": 0.2878,
+      "step": 4195
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.024011299435028e-05,
+      "loss": 0.4269,
+      "step": 4196
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0237758945386064e-05,
+      "loss": 0.2707,
+      "step": 4197
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.023540489642185e-05,
+      "loss": 0.2906,
+      "step": 4198
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.023305084745763e-05,
+      "loss": 0.2977,
+      "step": 4199
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.023069679849341e-05,
+      "loss": 0.4789,
+      "step": 4200
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.022834274952919e-05,
+      "loss": 0.3775,
+      "step": 4201
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.022598870056498e-05,
+      "loss": 0.2239,
+      "step": 4202
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0223634651600754e-05,
+      "loss": 0.2504,
+      "step": 4203
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.022128060263654e-05,
+      "loss": 0.2584,
+      "step": 4204
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.021892655367232e-05,
+      "loss": 0.2977,
+      "step": 4205
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.02165725047081e-05,
+      "loss": 0.2211,
+      "step": 4206
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0214218455743876e-05,
+      "loss": 0.3079,
+      "step": 4207
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.021186440677967e-05,
+      "loss": 0.3754,
+      "step": 4208
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0209510357815444e-05,
+      "loss": 0.4358,
+      "step": 4209
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.020715630885122e-05,
+      "loss": 0.4616,
+      "step": 4210
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.020480225988701e-05,
+      "loss": 0.3434,
+      "step": 4211
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.020244821092279e-05,
+      "loss": 0.2921,
+      "step": 4212
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0200094161958566e-05,
+      "loss": 0.4058,
+      "step": 4213
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.019774011299435e-05,
+      "loss": 0.4353,
+      "step": 4214
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0195386064030134e-05,
+      "loss": 0.2333,
+      "step": 4215
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.019303201506592e-05,
+      "loss": 0.3496,
+      "step": 4216
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0190677966101695e-05,
+      "loss": 0.2516,
+      "step": 4217
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.018832391713748e-05,
+      "loss": 0.285,
+      "step": 4218
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.018596986817326e-05,
+      "loss": 0.4167,
+      "step": 4219
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.018361581920904e-05,
+      "loss": 0.3132,
+      "step": 4220
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0181261770244824e-05,
+      "loss": 0.371,
+      "step": 4221
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.017890772128061e-05,
+      "loss": 0.2976,
+      "step": 4222
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0176553672316385e-05,
+      "loss": 0.3813,
+      "step": 4223
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.017419962335216e-05,
+      "loss": 0.4538,
+      "step": 4224
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.017184557438795e-05,
+      "loss": 0.2605,
+      "step": 4225
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.016949152542373e-05,
+      "loss": 0.2254,
+      "step": 4226
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.016713747645951e-05,
+      "loss": 0.5101,
+      "step": 4227
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.01647834274953e-05,
+      "loss": 0.3143,
+      "step": 4228
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0162429378531075e-05,
+      "loss": 0.3559,
+      "step": 4229
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.016007532956685e-05,
+      "loss": 0.3853,
+      "step": 4230
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0157721280602636e-05,
+      "loss": 0.2722,
+      "step": 4231
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.015536723163842e-05,
+      "loss": 0.2807,
+      "step": 4232
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0153013182674204e-05,
+      "loss": 0.2709,
+      "step": 4233
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.015065913370998e-05,
+      "loss": 0.2954,
+      "step": 4234
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0148305084745765e-05,
+      "loss": 0.3649,
+      "step": 4235
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.014595103578155e-05,
+      "loss": 0.3102,
+      "step": 4236
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0143596986817326e-05,
+      "loss": 0.4962,
+      "step": 4237
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.014124293785311e-05,
+      "loss": 0.3302,
+      "step": 4238
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0138888888888894e-05,
+      "loss": 0.2986,
+      "step": 4239
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.013653483992467e-05,
+      "loss": 0.2826,
+      "step": 4240
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.013418079096045e-05,
+      "loss": 0.3632,
+      "step": 4241
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.013182674199624e-05,
+      "loss": 0.3283,
+      "step": 4242
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0129472693032016e-05,
+      "loss": 0.4383,
+      "step": 4243
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.012711864406779e-05,
+      "loss": 0.2267,
+      "step": 4244
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0124764595103584e-05,
+      "loss": 0.2532,
+      "step": 4245
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.012241054613936e-05,
+      "loss": 0.2331,
+      "step": 4246
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0120056497175145e-05,
+      "loss": 0.2804,
+      "step": 4247
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.011770244821092e-05,
+      "loss": 0.2507,
+      "step": 4248
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.0115348399246706e-05,
+      "loss": 0.3725,
+      "step": 4249
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.011299435028249e-05,
+      "loss": 0.2826,
+      "step": 4250
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.011064030131827e-05,
+      "loss": 0.1912,
+      "step": 4251
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.010828625235405e-05,
+      "loss": 0.2806,
+      "step": 4252
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.0105932203389835e-05,
+      "loss": 0.2734,
+      "step": 4253
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.010357815442561e-05,
+      "loss": 0.3386,
+      "step": 4254
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.0101224105461396e-05,
+      "loss": 0.4069,
+      "step": 4255
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.009887005649718e-05,
+      "loss": 0.4076,
+      "step": 4256
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.009651600753296e-05,
+      "loss": 0.3156,
+      "step": 4257
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.009416195856874e-05,
+      "loss": 0.1351,
+      "step": 4258
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.8527589346019441,
+      "eval_loss": 0.35805782675743103,
+      "eval_runtime": 22.4157,
+      "eval_samples_per_second": 1519.155,
+      "eval_steps_per_second": 23.778,
+      "step": 4258
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.0091807909604525e-05,
+      "loss": 0.1437,
+      "step": 4259
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.00894538606403e-05,
+      "loss": 0.3206,
+      "step": 4260
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.008709981167608e-05,
+      "loss": 0.204,
+      "step": 4261
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.008474576271187e-05,
+      "loss": 0.2415,
+      "step": 4262
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.008239171374765e-05,
+      "loss": 0.0991,
+      "step": 4263
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.008003766478343e-05,
+      "loss": 0.1937,
+      "step": 4264
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.007768361581921e-05,
+      "loss": 0.1492,
+      "step": 4265
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.007532956685499e-05,
+      "loss": 0.2127,
+      "step": 4266
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.0072975517890776e-05,
+      "loss": 0.2807,
+      "step": 4267
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.007062146892655e-05,
+      "loss": 0.2165,
+      "step": 4268
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.006826741996234e-05,
+      "loss": 0.1917,
+      "step": 4269
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.006591337099812e-05,
+      "loss": 0.2583,
+      "step": 4270
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.00635593220339e-05,
+      "loss": 0.2052,
+      "step": 4271
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.006120527306968e-05,
+      "loss": 0.1917,
+      "step": 4272
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.0058851224105466e-05,
+      "loss": 0.1544,
+      "step": 4273
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.005649717514124e-05,
+      "loss": 0.4672,
+      "step": 4274
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.005414312617703e-05,
+      "loss": 0.1616,
+      "step": 4275
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.005178907721281e-05,
+      "loss": 0.257,
+      "step": 4276
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.004943502824859e-05,
+      "loss": 0.257,
+      "step": 4277
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.004708097928437e-05,
+      "loss": 0.2717,
+      "step": 4278
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.0044726930320156e-05,
+      "loss": 0.3498,
+      "step": 4279
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.004237288135593e-05,
+      "loss": 0.1946,
+      "step": 4280
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.004001883239172e-05,
+      "loss": 0.2643,
+      "step": 4281
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.0037664783427494e-05,
+      "loss": 0.1808,
+      "step": 4282
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.003531073446328e-05,
+      "loss": 0.2052,
+      "step": 4283
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.003295668549906e-05,
+      "loss": 0.2826,
+      "step": 4284
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.003060263653484e-05,
+      "loss": 0.1538,
+      "step": 4285
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.002824858757062e-05,
+      "loss": 0.1716,
+      "step": 4286
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.002589453860641e-05,
+      "loss": 0.2138,
+      "step": 4287
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.0023540489642184e-05,
+      "loss": 0.3316,
+      "step": 4288
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.002118644067797e-05,
+      "loss": 0.2465,
+      "step": 4289
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.001883239171375e-05,
+      "loss": 0.2007,
+      "step": 4290
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.001647834274953e-05,
+      "loss": 0.1576,
+      "step": 4291
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.001412429378531e-05,
+      "loss": 0.1634,
+      "step": 4292
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.00117702448211e-05,
+      "loss": 0.1677,
+      "step": 4293
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.0009416195856874e-05,
+      "loss": 0.2264,
+      "step": 4294
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.000706214689266e-05,
+      "loss": 0.1699,
+      "step": 4295
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.000470809792844e-05,
+      "loss": 0.2162,
+      "step": 4296
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.000235404896422e-05,
+      "loss": 0.3104,
+      "step": 4297
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4e-05,
+      "loss": 0.2641,
+      "step": 4298
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.999764595103578e-05,
+      "loss": 0.2528,
+      "step": 4299
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.9995291902071564e-05,
+      "loss": 0.1589,
+      "step": 4300
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.999293785310735e-05,
+      "loss": 0.1759,
+      "step": 4301
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.9990583804143125e-05,
+      "loss": 0.1687,
+      "step": 4302
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.998822975517891e-05,
+      "loss": 0.2365,
+      "step": 4303
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.998587570621469e-05,
+      "loss": 0.1708,
+      "step": 4304
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.998352165725047e-05,
+      "loss": 0.1516,
+      "step": 4305
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.9981167608286254e-05,
+      "loss": 0.3978,
+      "step": 4306
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.997881355932204e-05,
+      "loss": 0.2121,
+      "step": 4307
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.9976459510357815e-05,
+      "loss": 0.283,
+      "step": 4308
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.99741054613936e-05,
+      "loss": 0.103,
+      "step": 4309
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.997175141242938e-05,
+      "loss": 0.2994,
+      "step": 4310
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.996939736346516e-05,
+      "loss": 0.3164,
+      "step": 4311
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9967043314500944e-05,
+      "loss": 0.1586,
+      "step": 4312
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.996468926553673e-05,
+      "loss": 0.112,
+      "step": 4313
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9962335216572505e-05,
+      "loss": 0.4478,
+      "step": 4314
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.995998116760829e-05,
+      "loss": 0.1621,
+      "step": 4315
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9957627118644066e-05,
+      "loss": 0.1068,
+      "step": 4316
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.995527306967985e-05,
+      "loss": 0.1981,
+      "step": 4317
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9952919020715634e-05,
+      "loss": 0.3328,
+      "step": 4318
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.995056497175141e-05,
+      "loss": 0.1424,
+      "step": 4319
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9948210922787195e-05,
+      "loss": 0.2096,
+      "step": 4320
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.994585687382298e-05,
+      "loss": 0.2184,
+      "step": 4321
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9943502824858756e-05,
+      "loss": 0.125,
+      "step": 4322
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.994114877589454e-05,
+      "loss": 0.1826,
+      "step": 4323
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9938794726930324e-05,
+      "loss": 0.2048,
+      "step": 4324
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.99364406779661e-05,
+      "loss": 0.1341,
+      "step": 4325
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9934086629001885e-05,
+      "loss": 0.2707,
+      "step": 4326
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.993173258003767e-05,
+      "loss": 0.2141,
+      "step": 4327
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9929378531073446e-05,
+      "loss": 0.1407,
+      "step": 4328
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.992702448210923e-05,
+      "loss": 0.3573,
+      "step": 4329
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9924670433145014e-05,
+      "loss": 0.1151,
+      "step": 4330
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.992231638418079e-05,
+      "loss": 0.1958,
+      "step": 4331
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9919962335216575e-05,
+      "loss": 0.1277,
+      "step": 4332
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.991760828625236e-05,
+      "loss": 0.29,
+      "step": 4333
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9915254237288136e-05,
+      "loss": 0.1159,
+      "step": 4334
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.991290018832392e-05,
+      "loss": 0.1869,
+      "step": 4335
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.99105461393597e-05,
+      "loss": 0.2133,
+      "step": 4336
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.990819209039548e-05,
+      "loss": 0.2109,
+      "step": 4337
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9905838041431265e-05,
+      "loss": 0.2894,
+      "step": 4338
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.990348399246704e-05,
+      "loss": 0.1717,
+      "step": 4339
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9901129943502826e-05,
+      "loss": 0.1292,
+      "step": 4340
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.989877589453861e-05,
+      "loss": 0.1582,
+      "step": 4341
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.989642184557439e-05,
+      "loss": 0.0791,
+      "step": 4342
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.989406779661017e-05,
+      "loss": 0.1879,
+      "step": 4343
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9891713747645955e-05,
+      "loss": 0.2159,
+      "step": 4344
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.988935969868173e-05,
+      "loss": 0.1824,
+      "step": 4345
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9887005649717516e-05,
+      "loss": 0.1953,
+      "step": 4346
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.98846516007533e-05,
+      "loss": 0.208,
+      "step": 4347
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.988229755178908e-05,
+      "loss": 0.1944,
+      "step": 4348
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.987994350282486e-05,
+      "loss": 0.1018,
+      "step": 4349
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9877589453860645e-05,
+      "loss": 0.1397,
+      "step": 4350
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.987523540489642e-05,
+      "loss": 0.1519,
+      "step": 4351
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9872881355932206e-05,
+      "loss": 0.2743,
+      "step": 4352
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9870527306967984e-05,
+      "loss": 0.2528,
+      "step": 4353
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.986817325800377e-05,
+      "loss": 0.1428,
+      "step": 4354
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.986581920903955e-05,
+      "loss": 0.3111,
+      "step": 4355
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.986346516007533e-05,
+      "loss": 0.1516,
+      "step": 4356
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.986111111111111e-05,
+      "loss": 0.1345,
+      "step": 4357
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9858757062146896e-05,
+      "loss": 0.127,
+      "step": 4358
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9856403013182674e-05,
+      "loss": 0.4039,
+      "step": 4359
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.985404896421846e-05,
+      "loss": 0.4064,
+      "step": 4360
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.985169491525424e-05,
+      "loss": 0.2928,
+      "step": 4361
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.984934086629002e-05,
+      "loss": 0.2104,
+      "step": 4362
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.98469868173258e-05,
+      "loss": 0.2556,
+      "step": 4363
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9844632768361586e-05,
+      "loss": 0.2417,
+      "step": 4364
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9842278719397364e-05,
+      "loss": 0.1208,
+      "step": 4365
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.983992467043315e-05,
+      "loss": 0.0992,
+      "step": 4366
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.983757062146893e-05,
+      "loss": 0.1922,
+      "step": 4367
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.983521657250471e-05,
+      "loss": 0.3587,
+      "step": 4368
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.983286252354049e-05,
+      "loss": 0.303,
+      "step": 4369
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.983050847457627e-05,
+      "loss": 0.1411,
+      "step": 4370
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9828154425612054e-05,
+      "loss": 0.1937,
+      "step": 4371
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.982580037664784e-05,
+      "loss": 0.2117,
+      "step": 4372
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9823446327683615e-05,
+      "loss": 0.2398,
+      "step": 4373
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.98210922787194e-05,
+      "loss": 0.2255,
+      "step": 4374
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.981873822975518e-05,
+      "loss": 0.1571,
+      "step": 4375
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.981638418079096e-05,
+      "loss": 0.1507,
+      "step": 4376
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9814030131826744e-05,
+      "loss": 0.1304,
+      "step": 4377
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.981167608286253e-05,
+      "loss": 0.234,
+      "step": 4378
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9809322033898305e-05,
+      "loss": 0.1612,
+      "step": 4379
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.980696798493409e-05,
+      "loss": 0.1222,
+      "step": 4380
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.980461393596987e-05,
+      "loss": 0.1307,
+      "step": 4381
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.980225988700565e-05,
+      "loss": 0.1689,
+      "step": 4382
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9799905838041434e-05,
+      "loss": 0.2423,
+      "step": 4383
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.979755178907722e-05,
+      "loss": 0.2016,
+      "step": 4384
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9795197740112995e-05,
+      "loss": 0.3459,
+      "step": 4385
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.979284369114878e-05,
+      "loss": 0.1587,
+      "step": 4386
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9790489642184556e-05,
+      "loss": 0.1173,
+      "step": 4387
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.978813559322034e-05,
+      "loss": 0.1871,
+      "step": 4388
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9785781544256124e-05,
+      "loss": 0.1673,
+      "step": 4389
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.97834274952919e-05,
+      "loss": 0.1755,
+      "step": 4390
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9781073446327685e-05,
+      "loss": 0.3838,
+      "step": 4391
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.977871939736347e-05,
+      "loss": 0.1264,
+      "step": 4392
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9776365348399246e-05,
+      "loss": 0.2617,
+      "step": 4393
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.977401129943503e-05,
+      "loss": 0.2804,
+      "step": 4394
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9771657250470814e-05,
+      "loss": 0.2299,
+      "step": 4395
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.976930320150659e-05,
+      "loss": 0.2687,
+      "step": 4396
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9766949152542375e-05,
+      "loss": 0.2305,
+      "step": 4397
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.976459510357816e-05,
+      "loss": 0.135,
+      "step": 4398
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9762241054613936e-05,
+      "loss": 0.1375,
+      "step": 4399
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.975988700564972e-05,
+      "loss": 0.1401,
+      "step": 4400
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9757532956685504e-05,
+      "loss": 0.2488,
+      "step": 4401
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.975517890772128e-05,
+      "loss": 0.17,
+      "step": 4402
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9752824858757065e-05,
+      "loss": 0.1874,
+      "step": 4403
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.975047080979284e-05,
+      "loss": 0.1414,
+      "step": 4404
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.974811676082863e-05,
+      "loss": 0.1826,
+      "step": 4405
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.974576271186441e-05,
+      "loss": 0.2266,
+      "step": 4406
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.974340866290019e-05,
+      "loss": 0.2805,
+      "step": 4407
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.974105461393597e-05,
+      "loss": 0.2795,
+      "step": 4408
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9738700564971755e-05,
+      "loss": 0.2613,
+      "step": 4409
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.973634651600753e-05,
+      "loss": 0.2838,
+      "step": 4410
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9733992467043316e-05,
+      "loss": 0.1427,
+      "step": 4411
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.97316384180791e-05,
+      "loss": 0.1729,
+      "step": 4412
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.972928436911488e-05,
+      "loss": 0.2367,
+      "step": 4413
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.972693032015066e-05,
+      "loss": 0.2158,
+      "step": 4414
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9724576271186445e-05,
+      "loss": 0.1163,
+      "step": 4415
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.972222222222222e-05,
+      "loss": 0.1104,
+      "step": 4416
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9719868173258006e-05,
+      "loss": 0.2769,
+      "step": 4417
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.971751412429379e-05,
+      "loss": 0.1555,
+      "step": 4418
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.971516007532957e-05,
+      "loss": 0.3406,
+      "step": 4419
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.971280602636535e-05,
+      "loss": 0.2811,
+      "step": 4420
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.971045197740113e-05,
+      "loss": 0.2032,
+      "step": 4421
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.970809792843692e-05,
+      "loss": 0.19,
+      "step": 4422
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9705743879472696e-05,
+      "loss": 0.2327,
+      "step": 4423
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.970338983050847e-05,
+      "loss": 0.1668,
+      "step": 4424
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9701035781544264e-05,
+      "loss": 0.2282,
+      "step": 4425
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.969868173258004e-05,
+      "loss": 0.2339,
+      "step": 4426
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.969632768361582e-05,
+      "loss": 0.2519,
+      "step": 4427
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.96939736346516e-05,
+      "loss": 0.1311,
+      "step": 4428
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9691619585687386e-05,
+      "loss": 0.0977,
+      "step": 4429
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.968926553672316e-05,
+      "loss": 0.1798,
+      "step": 4430
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.968691148775895e-05,
+      "loss": 0.1447,
+      "step": 4431
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.968455743879473e-05,
+      "loss": 0.2197,
+      "step": 4432
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.968220338983051e-05,
+      "loss": 0.2428,
+      "step": 4433
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.967984934086629e-05,
+      "loss": 0.3245,
+      "step": 4434
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9677495291902076e-05,
+      "loss": 0.2536,
+      "step": 4435
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.967514124293786e-05,
+      "loss": 0.2461,
+      "step": 4436
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.967278719397364e-05,
+      "loss": 0.2454,
+      "step": 4437
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9670433145009414e-05,
+      "loss": 0.3479,
+      "step": 4438
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9668079096045205e-05,
+      "loss": 0.1821,
+      "step": 4439
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.966572504708098e-05,
+      "loss": 0.2355,
+      "step": 4440
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.966337099811676e-05,
+      "loss": 0.2296,
+      "step": 4441
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.966101694915255e-05,
+      "loss": 0.2019,
+      "step": 4442
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.965866290018833e-05,
+      "loss": 0.1286,
+      "step": 4443
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9656308851224104e-05,
+      "loss": 0.246,
+      "step": 4444
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.965395480225989e-05,
+      "loss": 0.2573,
+      "step": 4445
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.965160075329567e-05,
+      "loss": 0.2405,
+      "step": 4446
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.964924670433145e-05,
+      "loss": 0.2718,
+      "step": 4447
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.964689265536723e-05,
+      "loss": 0.2418,
+      "step": 4448
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.964453860640302e-05,
+      "loss": 0.2127,
+      "step": 4449
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9642184557438794e-05,
+      "loss": 0.1635,
+      "step": 4450
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.963983050847458e-05,
+      "loss": 0.1855,
+      "step": 4451
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.963747645951036e-05,
+      "loss": 0.2117,
+      "step": 4452
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9635122410546146e-05,
+      "loss": 0.2561,
+      "step": 4453
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.963276836158192e-05,
+      "loss": 0.2748,
+      "step": 4454
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.96304143126177e-05,
+      "loss": 0.1442,
+      "step": 4455
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.962806026365349e-05,
+      "loss": 0.1246,
+      "step": 4456
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.962570621468927e-05,
+      "loss": 0.2066,
+      "step": 4457
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9623352165725045e-05,
+      "loss": 0.1734,
+      "step": 4458
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9620998116760836e-05,
+      "loss": 0.2304,
+      "step": 4459
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.961864406779661e-05,
+      "loss": 0.3171,
+      "step": 4460
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.961629001883239e-05,
+      "loss": 0.2359,
+      "step": 4461
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9613935969868174e-05,
+      "loss": 0.1608,
+      "step": 4462
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.961158192090396e-05,
+      "loss": 0.1529,
+      "step": 4463
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9609227871939735e-05,
+      "loss": 0.1943,
+      "step": 4464
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.960687382297552e-05,
+      "loss": 0.2189,
+      "step": 4465
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.96045197740113e-05,
+      "loss": 0.24,
+      "step": 4466
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.960216572504708e-05,
+      "loss": 0.2195,
+      "step": 4467
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9599811676082864e-05,
+      "loss": 0.1866,
+      "step": 4468
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.959745762711865e-05,
+      "loss": 0.1049,
+      "step": 4469
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.959510357815443e-05,
+      "loss": 0.2259,
+      "step": 4470
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.959274952919021e-05,
+      "loss": 0.144,
+      "step": 4471
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9590395480225986e-05,
+      "loss": 0.1195,
+      "step": 4472
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.958804143126178e-05,
+      "loss": 0.138,
+      "step": 4473
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9585687382297554e-05,
+      "loss": 0.1949,
+      "step": 4474
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.958333333333333e-05,
+      "loss": 0.2192,
+      "step": 4475
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.958097928436912e-05,
+      "loss": 0.2078,
+      "step": 4476
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.95786252354049e-05,
+      "loss": 0.392,
+      "step": 4477
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9576271186440676e-05,
+      "loss": 0.1454,
+      "step": 4478
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.957391713747646e-05,
+      "loss": 0.1456,
+      "step": 4479
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9571563088512244e-05,
+      "loss": 0.2211,
+      "step": 4480
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.956920903954802e-05,
+      "loss": 0.1117,
+      "step": 4481
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9566854990583805e-05,
+      "loss": 0.075,
+      "step": 4482
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.956450094161959e-05,
+      "loss": 0.2414,
+      "step": 4483
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.956214689265537e-05,
+      "loss": 0.1894,
+      "step": 4484
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.955979284369115e-05,
+      "loss": 0.2468,
+      "step": 4485
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9557438794726934e-05,
+      "loss": 0.1813,
+      "step": 4486
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.955508474576272e-05,
+      "loss": 0.1744,
+      "step": 4487
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9552730696798495e-05,
+      "loss": 0.1805,
+      "step": 4488
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.955037664783427e-05,
+      "loss": 0.1743,
+      "step": 4489
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.954802259887006e-05,
+      "loss": 0.1328,
+      "step": 4490
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.954566854990584e-05,
+      "loss": 0.2543,
+      "step": 4491
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.954331450094162e-05,
+      "loss": 0.1401,
+      "step": 4492
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.954096045197741e-05,
+      "loss": 0.2521,
+      "step": 4493
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9538606403013185e-05,
+      "loss": 0.1802,
+      "step": 4494
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.953625235404896e-05,
+      "loss": 0.2104,
+      "step": 4495
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9533898305084746e-05,
+      "loss": 0.1866,
+      "step": 4496
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.953154425612053e-05,
+      "loss": 0.1771,
+      "step": 4497
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.952919020715631e-05,
+      "loss": 0.0754,
+      "step": 4498
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.952683615819209e-05,
+      "loss": 0.2171,
+      "step": 4499
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9524482109227875e-05,
+      "loss": 0.2365,
+      "step": 4500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.952212806026366e-05,
+      "loss": 0.1515,
+      "step": 4501
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9519774011299436e-05,
+      "loss": 0.2596,
+      "step": 4502
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.951741996233522e-05,
+      "loss": 0.2571,
+      "step": 4503
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9515065913371004e-05,
+      "loss": 0.1293,
+      "step": 4504
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.951271186440678e-05,
+      "loss": 0.2928,
+      "step": 4505
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.951035781544256e-05,
+      "loss": 0.2632,
+      "step": 4506
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.950800376647835e-05,
+      "loss": 0.2882,
+      "step": 4507
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9505649717514126e-05,
+      "loss": 0.4084,
+      "step": 4508
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.95032956685499e-05,
+      "loss": 0.2387,
+      "step": 4509
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9500941619585694e-05,
+      "loss": 0.2011,
+      "step": 4510
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.949858757062147e-05,
+      "loss": 0.1428,
+      "step": 4511
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.949623352165725e-05,
+      "loss": 0.2496,
+      "step": 4512
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.949387947269303e-05,
+      "loss": 0.2505,
+      "step": 4513
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9491525423728816e-05,
+      "loss": 0.2781,
+      "step": 4514
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.94891713747646e-05,
+      "loss": 0.1618,
+      "step": 4515
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.948681732580038e-05,
+      "loss": 0.1399,
+      "step": 4516
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.948446327683616e-05,
+      "loss": 0.1323,
+      "step": 4517
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9482109227871945e-05,
+      "loss": 0.2159,
+      "step": 4518
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.947975517890772e-05,
+      "loss": 0.1708,
+      "step": 4519
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9477401129943506e-05,
+      "loss": 0.2196,
+      "step": 4520
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.947504708097929e-05,
+      "loss": 0.3254,
+      "step": 4521
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.947269303201507e-05,
+      "loss": 0.2031,
+      "step": 4522
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9470338983050844e-05,
+      "loss": 0.1429,
+      "step": 4523
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9467984934086635e-05,
+      "loss": 0.2895,
+      "step": 4524
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.946563088512241e-05,
+      "loss": 0.1207,
+      "step": 4525
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.946327683615819e-05,
+      "loss": 0.2096,
+      "step": 4526
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.946092278719398e-05,
+      "loss": 0.4196,
+      "step": 4527
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.945856873822976e-05,
+      "loss": 0.2388,
+      "step": 4528
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.9456214689265534e-05,
+      "loss": 0.2754,
+      "step": 4529
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.945386064030132e-05,
+      "loss": 0.2532,
+      "step": 4530
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.94515065913371e-05,
+      "loss": 0.4175,
+      "step": 4531
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.9449152542372886e-05,
+      "loss": 0.3124,
+      "step": 4532
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.944679849340866e-05,
+      "loss": 0.1657,
+      "step": 4533
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.944444444444445e-05,
+      "loss": 0.1781,
+      "step": 4534
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.944209039548023e-05,
+      "loss": 0.4341,
+      "step": 4535
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.943973634651601e-05,
+      "loss": 0.161,
+      "step": 4536
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.943738229755179e-05,
+      "loss": 0.2274,
+      "step": 4537
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.9435028248587576e-05,
+      "loss": 0.2072,
+      "step": 4538
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.943267419962335e-05,
+      "loss": 0.3158,
+      "step": 4539
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.943032015065913e-05,
+      "loss": 0.1781,
+      "step": 4540
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.942796610169492e-05,
+      "loss": 0.1914,
+      "step": 4541
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.94256120527307e-05,
+      "loss": 0.225,
+      "step": 4542
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.9423258003766475e-05,
+      "loss": 0.2695,
+      "step": 4543
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.9420903954802266e-05,
+      "loss": 0.1333,
+      "step": 4544
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.941854990583804e-05,
+      "loss": 0.253,
+      "step": 4545
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.941619585687382e-05,
+      "loss": 0.2109,
+      "step": 4546
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9413841807909604e-05,
+      "loss": 0.2654,
+      "step": 4547
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.941148775894539e-05,
+      "loss": 0.1951,
+      "step": 4548
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.940913370998117e-05,
+      "loss": 0.3021,
+      "step": 4549
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.940677966101695e-05,
+      "loss": 0.3308,
+      "step": 4550
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.940442561205273e-05,
+      "loss": 0.1999,
+      "step": 4551
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.940207156308852e-05,
+      "loss": 0.2409,
+      "step": 4552
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9399717514124294e-05,
+      "loss": 0.2736,
+      "step": 4553
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.939736346516008e-05,
+      "loss": 0.293,
+      "step": 4554
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.939500941619586e-05,
+      "loss": 0.2175,
+      "step": 4555
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.939265536723164e-05,
+      "loss": 0.3302,
+      "step": 4556
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9390301318267416e-05,
+      "loss": 0.2954,
+      "step": 4557
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.938794726930321e-05,
+      "loss": 0.167,
+      "step": 4558
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9385593220338984e-05,
+      "loss": 0.1687,
+      "step": 4559
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.938323917137476e-05,
+      "loss": 0.2803,
+      "step": 4560
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.938088512241055e-05,
+      "loss": 0.2199,
+      "step": 4561
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.937853107344633e-05,
+      "loss": 0.211,
+      "step": 4562
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.937617702448211e-05,
+      "loss": 0.2144,
+      "step": 4563
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.937382297551789e-05,
+      "loss": 0.2608,
+      "step": 4564
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9371468926553674e-05,
+      "loss": 0.2347,
+      "step": 4565
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.936911487758946e-05,
+      "loss": 0.2185,
+      "step": 4566
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9366760828625235e-05,
+      "loss": 0.179,
+      "step": 4567
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.936440677966102e-05,
+      "loss": 0.302,
+      "step": 4568
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.93620527306968e-05,
+      "loss": 0.2545,
+      "step": 4569
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.935969868173258e-05,
+      "loss": 0.2702,
+      "step": 4570
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9357344632768364e-05,
+      "loss": 0.2259,
+      "step": 4571
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.935499058380415e-05,
+      "loss": 0.383,
+      "step": 4572
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9352636534839925e-05,
+      "loss": 0.2263,
+      "step": 4573
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.935028248587571e-05,
+      "loss": 0.1747,
+      "step": 4574
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.934792843691149e-05,
+      "loss": 0.1325,
+      "step": 4575
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.934557438794727e-05,
+      "loss": 0.3585,
+      "step": 4576
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.934322033898305e-05,
+      "loss": 0.2097,
+      "step": 4577
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.934086629001884e-05,
+      "loss": 0.2613,
+      "step": 4578
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9338512241054615e-05,
+      "loss": 0.2214,
+      "step": 4579
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.93361581920904e-05,
+      "loss": 0.249,
+      "step": 4580
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9333804143126176e-05,
+      "loss": 0.1516,
+      "step": 4581
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.933145009416196e-05,
+      "loss": 0.2025,
+      "step": 4582
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9329096045197744e-05,
+      "loss": 0.2099,
+      "step": 4583
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.932674199623352e-05,
+      "loss": 0.2055,
+      "step": 4584
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9324387947269305e-05,
+      "loss": 0.3422,
+      "step": 4585
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.932203389830509e-05,
+      "loss": 0.2301,
+      "step": 4586
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9319679849340866e-05,
+      "loss": 0.2989,
+      "step": 4587
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.931732580037665e-05,
+      "loss": 0.1415,
+      "step": 4588
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9314971751412434e-05,
+      "loss": 0.2094,
+      "step": 4589
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.931261770244821e-05,
+      "loss": 0.21,
+      "step": 4590
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9310263653483995e-05,
+      "loss": 0.1775,
+      "step": 4591
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.930790960451978e-05,
+      "loss": 0.1278,
+      "step": 4592
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9305555555555556e-05,
+      "loss": 0.2285,
+      "step": 4593
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.930320150659134e-05,
+      "loss": 0.2219,
+      "step": 4594
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9300847457627124e-05,
+      "loss": 0.17,
+      "step": 4595
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.92984934086629e-05,
+      "loss": 0.1854,
+      "step": 4596
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9296139359698685e-05,
+      "loss": 0.1703,
+      "step": 4597
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.929378531073446e-05,
+      "loss": 0.124,
+      "step": 4598
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9291431261770246e-05,
+      "loss": 0.2745,
+      "step": 4599
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.928907721280603e-05,
+      "loss": 0.1721,
+      "step": 4600
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.928672316384181e-05,
+      "loss": 0.2634,
+      "step": 4601
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.928436911487759e-05,
+      "loss": 0.2052,
+      "step": 4602
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9282015065913375e-05,
+      "loss": 0.1879,
+      "step": 4603
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.927966101694915e-05,
+      "loss": 0.2125,
+      "step": 4604
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9277306967984936e-05,
+      "loss": 0.2576,
+      "step": 4605
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.927495291902072e-05,
+      "loss": 0.2729,
+      "step": 4606
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.92725988700565e-05,
+      "loss": 0.1792,
+      "step": 4607
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.927024482109228e-05,
+      "loss": 0.2366,
+      "step": 4608
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9267890772128065e-05,
+      "loss": 0.3411,
+      "step": 4609
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.926553672316384e-05,
+      "loss": 0.1089,
+      "step": 4610
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9263182674199626e-05,
+      "loss": 0.1971,
+      "step": 4611
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.926082862523541e-05,
+      "loss": 0.2533,
+      "step": 4612
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.925847457627119e-05,
+      "loss": 0.2161,
+      "step": 4613
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.925612052730697e-05,
+      "loss": 0.3136,
+      "step": 4614
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.925376647834275e-05,
+      "loss": 0.2455,
+      "step": 4615
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.925141242937853e-05,
+      "loss": 0.1772,
+      "step": 4616
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9249058380414316e-05,
+      "loss": 0.2003,
+      "step": 4617
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.924670433145009e-05,
+      "loss": 0.2758,
+      "step": 4618
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.924435028248588e-05,
+      "loss": 0.2006,
+      "step": 4619
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.924199623352166e-05,
+      "loss": 0.1413,
+      "step": 4620
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.923964218455744e-05,
+      "loss": 0.1815,
+      "step": 4621
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.923728813559322e-05,
+      "loss": 0.2784,
+      "step": 4622
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9234934086629006e-05,
+      "loss": 0.3812,
+      "step": 4623
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.923258003766478e-05,
+      "loss": 0.242,
+      "step": 4624
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.923022598870057e-05,
+      "loss": 0.168,
+      "step": 4625
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.922787193973635e-05,
+      "loss": 0.1402,
+      "step": 4626
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.922551789077213e-05,
+      "loss": 0.135,
+      "step": 4627
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.922316384180791e-05,
+      "loss": 0.3794,
+      "step": 4628
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9220809792843696e-05,
+      "loss": 0.2658,
+      "step": 4629
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.921845574387947e-05,
+      "loss": 0.3352,
+      "step": 4630
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.921610169491526e-05,
+      "loss": 0.2713,
+      "step": 4631
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.9213747645951034e-05,
+      "loss": 0.2413,
+      "step": 4632
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.921139359698682e-05,
+      "loss": 0.2832,
+      "step": 4633
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.92090395480226e-05,
+      "loss": 0.1507,
+      "step": 4634
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.920668549905838e-05,
+      "loss": 0.2071,
+      "step": 4635
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.920433145009416e-05,
+      "loss": 0.187,
+      "step": 4636
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.920197740112995e-05,
+      "loss": 0.1174,
+      "step": 4637
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.9199623352165724e-05,
+      "loss": 0.3248,
+      "step": 4638
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.919726930320151e-05,
+      "loss": 0.165,
+      "step": 4639
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.919491525423729e-05,
+      "loss": 0.1629,
+      "step": 4640
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.919256120527307e-05,
+      "loss": 0.2864,
+      "step": 4641
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.919020715630885e-05,
+      "loss": 0.2255,
+      "step": 4642
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.918785310734464e-05,
+      "loss": 0.1436,
+      "step": 4643
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.9185499058380414e-05,
+      "loss": 0.1406,
+      "step": 4644
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.91831450094162e-05,
+      "loss": 0.258,
+      "step": 4645
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.918079096045198e-05,
+      "loss": 0.1905,
+      "step": 4646
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.917843691148776e-05,
+      "loss": 0.2441,
+      "step": 4647
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.917608286252354e-05,
+      "loss": 0.173,
+      "step": 4648
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.917372881355933e-05,
+      "loss": 0.2843,
+      "step": 4649
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.9171374764595104e-05,
+      "loss": 0.2413,
+      "step": 4650
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.916902071563089e-05,
+      "loss": 0.2074,
+      "step": 4651
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9166666666666665e-05,
+      "loss": 0.2513,
+      "step": 4652
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.916431261770245e-05,
+      "loss": 0.138,
+      "step": 4653
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.916195856873823e-05,
+      "loss": 0.2031,
+      "step": 4654
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.915960451977401e-05,
+      "loss": 0.1981,
+      "step": 4655
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9157250470809794e-05,
+      "loss": 0.1728,
+      "step": 4656
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.915489642184558e-05,
+      "loss": 0.278,
+      "step": 4657
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9152542372881355e-05,
+      "loss": 0.2942,
+      "step": 4658
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.915018832391714e-05,
+      "loss": 0.2944,
+      "step": 4659
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.914783427495292e-05,
+      "loss": 0.1219,
+      "step": 4660
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.91454802259887e-05,
+      "loss": 0.289,
+      "step": 4661
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9143126177024484e-05,
+      "loss": 0.2902,
+      "step": 4662
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.914077212806027e-05,
+      "loss": 0.1047,
+      "step": 4663
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9138418079096045e-05,
+      "loss": 0.1936,
+      "step": 4664
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.913606403013183e-05,
+      "loss": 0.0857,
+      "step": 4665
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.913370998116761e-05,
+      "loss": 0.2457,
+      "step": 4666
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.913135593220339e-05,
+      "loss": 0.1362,
+      "step": 4667
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9129001883239174e-05,
+      "loss": 0.2183,
+      "step": 4668
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.912664783427495e-05,
+      "loss": 0.1591,
+      "step": 4669
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9124293785310735e-05,
+      "loss": 0.1672,
+      "step": 4670
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.912193973634652e-05,
+      "loss": 0.1184,
+      "step": 4671
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9119585687382297e-05,
+      "loss": 0.158,
+      "step": 4672
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.911723163841808e-05,
+      "loss": 0.2409,
+      "step": 4673
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.9114877589453864e-05,
+      "loss": 0.2342,
+      "step": 4674
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.911252354048964e-05,
+      "loss": 0.2972,
+      "step": 4675
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.9110169491525425e-05,
+      "loss": 0.2885,
+      "step": 4676
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.910781544256121e-05,
+      "loss": 0.2343,
+      "step": 4677
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.9105461393596987e-05,
+      "loss": 0.2115,
+      "step": 4678
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.910310734463277e-05,
+      "loss": 0.1357,
+      "step": 4679
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.9100753295668554e-05,
+      "loss": 0.2621,
+      "step": 4680
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.909839924670433e-05,
+      "loss": 0.2531,
+      "step": 4681
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.9096045197740115e-05,
+      "loss": 0.3213,
+      "step": 4682
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.90936911487759e-05,
+      "loss": 0.2049,
+      "step": 4683
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.9091337099811677e-05,
+      "loss": 0.2709,
+      "step": 4684
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.908898305084746e-05,
+      "loss": 0.147,
+      "step": 4685
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.908662900188324e-05,
+      "loss": 0.2216,
+      "step": 4686
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.908427495291902e-05,
+      "loss": 0.2008,
+      "step": 4687
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.9081920903954805e-05,
+      "loss": 0.173,
+      "step": 4688
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.907956685499058e-05,
+      "loss": 0.1989,
+      "step": 4689
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.9077212806026367e-05,
+      "loss": 0.1836,
+      "step": 4690
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.907485875706215e-05,
+      "loss": 0.1439,
+      "step": 4691
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.907250470809793e-05,
+      "loss": 0.1454,
+      "step": 4692
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.907015065913371e-05,
+      "loss": 0.2183,
+      "step": 4693
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.9067796610169495e-05,
+      "loss": 0.3539,
+      "step": 4694
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.906544256120527e-05,
+      "loss": 0.2204,
+      "step": 4695
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.9063088512241057e-05,
+      "loss": 0.1837,
+      "step": 4696
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.906073446327684e-05,
+      "loss": 0.1584,
+      "step": 4697
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.905838041431262e-05,
+      "loss": 0.2323,
+      "step": 4698
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.90560263653484e-05,
+      "loss": 0.2495,
+      "step": 4699
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.9053672316384185e-05,
+      "loss": 0.177,
+      "step": 4700
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.905131826741996e-05,
+      "loss": 0.1239,
+      "step": 4701
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.9048964218455747e-05,
+      "loss": 0.1068,
+      "step": 4702
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.9046610169491524e-05,
+      "loss": 0.192,
+      "step": 4703
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.904425612052731e-05,
+      "loss": 0.1787,
+      "step": 4704
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.904190207156309e-05,
+      "loss": 0.1833,
+      "step": 4705
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.903954802259887e-05,
+      "loss": 0.3296,
+      "step": 4706
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.903719397363465e-05,
+      "loss": 0.1687,
+      "step": 4707
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.9034839924670437e-05,
+      "loss": 0.1778,
+      "step": 4708
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.9032485875706214e-05,
+      "loss": 0.2401,
+      "step": 4709
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.9030131826742e-05,
+      "loss": 0.1377,
+      "step": 4710
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.902777777777778e-05,
+      "loss": 0.3079,
+      "step": 4711
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.902542372881356e-05,
+      "loss": 0.3309,
+      "step": 4712
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.902306967984934e-05,
+      "loss": 0.266,
+      "step": 4713
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.9020715630885127e-05,
+      "loss": 0.2167,
+      "step": 4714
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.9018361581920904e-05,
+      "loss": 0.181,
+      "step": 4715
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.901600753295669e-05,
+      "loss": 0.1842,
+      "step": 4716
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.901365348399247e-05,
+      "loss": 0.249,
+      "step": 4717
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.901129943502825e-05,
+      "loss": 0.2012,
+      "step": 4718
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.900894538606403e-05,
+      "loss": 0.3728,
+      "step": 4719
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.900659133709981e-05,
+      "loss": 0.2595,
+      "step": 4720
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.90042372881356e-05,
+      "loss": 0.1483,
+      "step": 4721
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.900188323917138e-05,
+      "loss": 0.1435,
+      "step": 4722
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.8999529190207155e-05,
+      "loss": 0.2635,
+      "step": 4723
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.899717514124294e-05,
+      "loss": 0.1454,
+      "step": 4724
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.899482109227872e-05,
+      "loss": 0.165,
+      "step": 4725
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.89924670433145e-05,
+      "loss": 0.2094,
+      "step": 4726
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.8990112994350284e-05,
+      "loss": 0.259,
+      "step": 4727
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.898775894538607e-05,
+      "loss": 0.1272,
+      "step": 4728
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.8985404896421845e-05,
+      "loss": 0.3284,
+      "step": 4729
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.898305084745763e-05,
+      "loss": 0.2108,
+      "step": 4730
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.898069679849341e-05,
+      "loss": 0.2582,
+      "step": 4731
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.897834274952919e-05,
+      "loss": 0.1901,
+      "step": 4732
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.8975988700564974e-05,
+      "loss": 0.2418,
+      "step": 4733
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.897363465160076e-05,
+      "loss": 0.242,
+      "step": 4734
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.8971280602636535e-05,
+      "loss": 0.199,
+      "step": 4735
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.896892655367232e-05,
+      "loss": 0.1963,
+      "step": 4736
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.8966572504708096e-05,
+      "loss": 0.0527,
+      "step": 4737
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8964218455743886e-05,
+      "loss": 0.2788,
+      "step": 4738
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8961864406779664e-05,
+      "loss": 0.4015,
+      "step": 4739
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.895951035781544e-05,
+      "loss": 0.1779,
+      "step": 4740
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.895715630885123e-05,
+      "loss": 0.181,
+      "step": 4741
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.895480225988701e-05,
+      "loss": 0.2021,
+      "step": 4742
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8952448210922786e-05,
+      "loss": 0.313,
+      "step": 4743
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.895009416195857e-05,
+      "loss": 0.3547,
+      "step": 4744
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8947740112994354e-05,
+      "loss": 0.2226,
+      "step": 4745
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.894538606403013e-05,
+      "loss": 0.2177,
+      "step": 4746
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8943032015065915e-05,
+      "loss": 0.2187,
+      "step": 4747
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.89406779661017e-05,
+      "loss": 0.2174,
+      "step": 4748
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8938323917137476e-05,
+      "loss": 0.2093,
+      "step": 4749
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.893596986817326e-05,
+      "loss": 0.2555,
+      "step": 4750
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8933615819209044e-05,
+      "loss": 0.2091,
+      "step": 4751
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.893126177024483e-05,
+      "loss": 0.2218,
+      "step": 4752
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8928907721280605e-05,
+      "loss": 0.1413,
+      "step": 4753
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.892655367231638e-05,
+      "loss": 0.3278,
+      "step": 4754
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.892419962335217e-05,
+      "loss": 0.3012,
+      "step": 4755
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.892184557438795e-05,
+      "loss": 0.4715,
+      "step": 4756
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.891949152542373e-05,
+      "loss": 0.206,
+      "step": 4757
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.891713747645952e-05,
+      "loss": 0.2315,
+      "step": 4758
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.8914783427495295e-05,
+      "loss": 0.2394,
+      "step": 4759
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.891242937853107e-05,
+      "loss": 0.3403,
+      "step": 4760
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.8910075329566856e-05,
+      "loss": 0.1245,
+      "step": 4761
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.890772128060264e-05,
+      "loss": 0.1328,
+      "step": 4762
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.890536723163842e-05,
+      "loss": 0.3028,
+      "step": 4763
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.89030131826742e-05,
+      "loss": 0.1572,
+      "step": 4764
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.8900659133709985e-05,
+      "loss": 0.2326,
+      "step": 4765
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.889830508474576e-05,
+      "loss": 0.2498,
+      "step": 4766
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.8895951035781546e-05,
+      "loss": 0.2452,
+      "step": 4767
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.889359698681733e-05,
+      "loss": 0.2264,
+      "step": 4768
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.8891242937853114e-05,
+      "loss": 0.2455,
+      "step": 4769
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.888888888888889e-05,
+      "loss": 0.1584,
+      "step": 4770
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.888653483992467e-05,
+      "loss": 0.2839,
+      "step": 4771
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.888418079096046e-05,
+      "loss": 0.1292,
+      "step": 4772
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.8881826741996236e-05,
+      "loss": 0.2395,
+      "step": 4773
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.887947269303201e-05,
+      "loss": 0.2673,
+      "step": 4774
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.8877118644067804e-05,
+      "loss": 0.2554,
+      "step": 4775
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.887476459510358e-05,
+      "loss": 0.2661,
+      "step": 4776
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.887241054613936e-05,
+      "loss": 0.2482,
+      "step": 4777
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.887005649717514e-05,
+      "loss": 0.2616,
+      "step": 4778
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.8867702448210926e-05,
+      "loss": 0.3205,
+      "step": 4779
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.88653483992467e-05,
+      "loss": 0.1586,
+      "step": 4780
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.886299435028249e-05,
+      "loss": 0.1388,
+      "step": 4781
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.886064030131827e-05,
+      "loss": 0.1466,
+      "step": 4782
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.885828625235405e-05,
+      "loss": 0.2043,
+      "step": 4783
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.885593220338983e-05,
+      "loss": 0.2896,
+      "step": 4784
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8853578154425616e-05,
+      "loss": 0.2573,
+      "step": 4785
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.88512241054614e-05,
+      "loss": 0.1429,
+      "step": 4786
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.884887005649718e-05,
+      "loss": 0.3062,
+      "step": 4787
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8846516007532954e-05,
+      "loss": 0.1705,
+      "step": 4788
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8844161958568745e-05,
+      "loss": 0.4141,
+      "step": 4789
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.884180790960452e-05,
+      "loss": 0.1801,
+      "step": 4790
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.88394538606403e-05,
+      "loss": 0.2247,
+      "step": 4791
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.883709981167609e-05,
+      "loss": 0.1572,
+      "step": 4792
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.883474576271187e-05,
+      "loss": 0.18,
+      "step": 4793
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8832391713747644e-05,
+      "loss": 0.0959,
+      "step": 4794
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.883003766478343e-05,
+      "loss": 0.3375,
+      "step": 4795
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.882768361581921e-05,
+      "loss": 0.2696,
+      "step": 4796
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.882532956685499e-05,
+      "loss": 0.1463,
+      "step": 4797
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.882297551789077e-05,
+      "loss": 0.2137,
+      "step": 4798
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.882062146892656e-05,
+      "loss": 0.1835,
+      "step": 4799
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.881826741996234e-05,
+      "loss": 0.1725,
+      "step": 4800
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.881591337099812e-05,
+      "loss": 0.2558,
+      "step": 4801
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.88135593220339e-05,
+      "loss": 0.3175,
+      "step": 4802
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8811205273069686e-05,
+      "loss": 0.2812,
+      "step": 4803
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.880885122410546e-05,
+      "loss": 0.1756,
+      "step": 4804
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.880649717514124e-05,
+      "loss": 0.2904,
+      "step": 4805
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.880414312617703e-05,
+      "loss": 0.1744,
+      "step": 4806
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.880178907721281e-05,
+      "loss": 0.1635,
+      "step": 4807
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8799435028248585e-05,
+      "loss": 0.2576,
+      "step": 4808
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8797080979284376e-05,
+      "loss": 0.3393,
+      "step": 4809
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.879472693032015e-05,
+      "loss": 0.1641,
+      "step": 4810
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.879237288135593e-05,
+      "loss": 0.1694,
+      "step": 4811
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8790018832391714e-05,
+      "loss": 0.0835,
+      "step": 4812
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.87876647834275e-05,
+      "loss": 0.2685,
+      "step": 4813
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8785310734463275e-05,
+      "loss": 0.2071,
+      "step": 4814
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.878295668549906e-05,
+      "loss": 0.323,
+      "step": 4815
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.878060263653484e-05,
+      "loss": 0.1552,
+      "step": 4816
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.877824858757063e-05,
+      "loss": 0.176,
+      "step": 4817
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8775894538606404e-05,
+      "loss": 0.2181,
+      "step": 4818
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.877354048964219e-05,
+      "loss": 0.2088,
+      "step": 4819
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.877118644067797e-05,
+      "loss": 0.2153,
+      "step": 4820
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.876883239171375e-05,
+      "loss": 0.3275,
+      "step": 4821
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8766478342749526e-05,
+      "loss": 0.148,
+      "step": 4822
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.876412429378532e-05,
+      "loss": 0.2455,
+      "step": 4823
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.8761770244821094e-05,
+      "loss": 0.2384,
+      "step": 4824
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.875941619585687e-05,
+      "loss": 0.221,
+      "step": 4825
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.875706214689266e-05,
+      "loss": 0.1732,
+      "step": 4826
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.875470809792844e-05,
+      "loss": 0.1235,
+      "step": 4827
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.8752354048964216e-05,
+      "loss": 0.0869,
+      "step": 4828
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.875e-05,
+      "loss": 0.187,
+      "step": 4829
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.8747645951035784e-05,
+      "loss": 0.2955,
+      "step": 4830
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.874529190207157e-05,
+      "loss": 0.191,
+      "step": 4831
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.8742937853107345e-05,
+      "loss": 0.2269,
+      "step": 4832
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.874058380414313e-05,
+      "loss": 0.2405,
+      "step": 4833
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.873822975517891e-05,
+      "loss": 0.2015,
+      "step": 4834
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.873587570621469e-05,
+      "loss": 0.2866,
+      "step": 4835
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.8733521657250474e-05,
+      "loss": 0.1957,
+      "step": 4836
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.873116760828626e-05,
+      "loss": 0.2761,
+      "step": 4837
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.8728813559322035e-05,
+      "loss": 0.3913,
+      "step": 4838
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.872645951035781e-05,
+      "loss": 0.1914,
+      "step": 4839
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.87241054613936e-05,
+      "loss": 0.1806,
+      "step": 4840
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.872175141242938e-05,
+      "loss": 0.3302,
+      "step": 4841
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.871939736346516e-05,
+      "loss": 0.161,
+      "step": 4842
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.871704331450095e-05,
+      "loss": 0.2224,
+      "step": 4843
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8714689265536725e-05,
+      "loss": 0.1198,
+      "step": 4844
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.87123352165725e-05,
+      "loss": 0.213,
+      "step": 4845
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8709981167608286e-05,
+      "loss": 0.2147,
+      "step": 4846
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.870762711864407e-05,
+      "loss": 0.1837,
+      "step": 4847
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8705273069679854e-05,
+      "loss": 0.1232,
+      "step": 4848
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.870291902071563e-05,
+      "loss": 0.1789,
+      "step": 4849
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8700564971751415e-05,
+      "loss": 0.2508,
+      "step": 4850
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.86982109227872e-05,
+      "loss": 0.2099,
+      "step": 4851
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8695856873822976e-05,
+      "loss": 0.2579,
+      "step": 4852
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.869350282485876e-05,
+      "loss": 0.1173,
+      "step": 4853
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8691148775894544e-05,
+      "loss": 0.1482,
+      "step": 4854
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.868879472693032e-05,
+      "loss": 0.2622,
+      "step": 4855
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.86864406779661e-05,
+      "loss": 0.2637,
+      "step": 4856
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.868408662900189e-05,
+      "loss": 0.2618,
+      "step": 4857
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8681732580037666e-05,
+      "loss": 0.1998,
+      "step": 4858
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.867937853107344e-05,
+      "loss": 0.2938,
+      "step": 4859
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8677024482109234e-05,
+      "loss": 0.212,
+      "step": 4860
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.867467043314501e-05,
+      "loss": 0.3069,
+      "step": 4861
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.867231638418079e-05,
+      "loss": 0.1799,
+      "step": 4862
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.866996233521657e-05,
+      "loss": 0.1624,
+      "step": 4863
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8667608286252356e-05,
+      "loss": 0.1192,
+      "step": 4864
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.866525423728814e-05,
+      "loss": 0.1646,
+      "step": 4865
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.866290018832392e-05,
+      "loss": 0.1152,
+      "step": 4866
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.86605461393597e-05,
+      "loss": 0.2823,
+      "step": 4867
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.8658192090395485e-05,
+      "loss": 0.1992,
+      "step": 4868
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.865583804143126e-05,
+      "loss": 0.2093,
+      "step": 4869
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.8653483992467046e-05,
+      "loss": 0.1789,
+      "step": 4870
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.865112994350283e-05,
+      "loss": 0.2422,
+      "step": 4871
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.864877589453861e-05,
+      "loss": 0.2291,
+      "step": 4872
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.8646421845574384e-05,
+      "loss": 0.1818,
+      "step": 4873
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.8644067796610175e-05,
+      "loss": 0.2884,
+      "step": 4874
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.864171374764595e-05,
+      "loss": 0.215,
+      "step": 4875
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.863935969868173e-05,
+      "loss": 0.2567,
+      "step": 4876
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.863700564971752e-05,
+      "loss": 0.1043,
+      "step": 4877
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.86346516007533e-05,
+      "loss": 0.1273,
+      "step": 4878
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.863229755178908e-05,
+      "loss": 0.3578,
+      "step": 4879
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.862994350282486e-05,
+      "loss": 0.16,
+      "step": 4880
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.862758945386064e-05,
+      "loss": 0.2941,
+      "step": 4881
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.8625235404896426e-05,
+      "loss": 0.2543,
+      "step": 4882
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.86228813559322e-05,
+      "loss": 0.3545,
+      "step": 4883
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.862052730696799e-05,
+      "loss": 0.2269,
+      "step": 4884
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.861817325800377e-05,
+      "loss": 0.3548,
+      "step": 4885
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.861581920903955e-05,
+      "loss": 0.2444,
+      "step": 4886
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.861346516007533e-05,
+      "loss": 0.1557,
+      "step": 4887
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.8611111111111116e-05,
+      "loss": 0.1582,
+      "step": 4888
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.860875706214689e-05,
+      "loss": 0.3274,
+      "step": 4889
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.860640301318268e-05,
+      "loss": 0.1809,
+      "step": 4890
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.860404896421846e-05,
+      "loss": 0.2809,
+      "step": 4891
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.860169491525424e-05,
+      "loss": 0.0753,
+      "step": 4892
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.8599340866290015e-05,
+      "loss": 0.2189,
+      "step": 4893
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.8596986817325806e-05,
+      "loss": 0.2128,
+      "step": 4894
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.859463276836158e-05,
+      "loss": 0.338,
+      "step": 4895
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.859227871939737e-05,
+      "loss": 0.3718,
+      "step": 4896
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.8589924670433144e-05,
+      "loss": 0.1663,
+      "step": 4897
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.858757062146893e-05,
+      "loss": 0.1758,
+      "step": 4898
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.858521657250471e-05,
+      "loss": 0.2832,
+      "step": 4899
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.858286252354049e-05,
+      "loss": 0.2558,
+      "step": 4900
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.858050847457627e-05,
+      "loss": 0.1363,
+      "step": 4901
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.857815442561206e-05,
+      "loss": 0.1627,
+      "step": 4902
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.8575800376647834e-05,
+      "loss": 0.1379,
+      "step": 4903
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.857344632768362e-05,
+      "loss": 0.2796,
+      "step": 4904
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.85710922787194e-05,
+      "loss": 0.1999,
+      "step": 4905
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.856873822975518e-05,
+      "loss": 0.188,
+      "step": 4906
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.856638418079096e-05,
+      "loss": 0.3445,
+      "step": 4907
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.856403013182675e-05,
+      "loss": 0.2091,
+      "step": 4908
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.8561676082862524e-05,
+      "loss": 0.1827,
+      "step": 4909
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.855932203389831e-05,
+      "loss": 0.2474,
+      "step": 4910
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.855696798493409e-05,
+      "loss": 0.254,
+      "step": 4911
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.855461393596987e-05,
+      "loss": 0.1428,
+      "step": 4912
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.855225988700565e-05,
+      "loss": 0.2039,
+      "step": 4913
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.854990583804143e-05,
+      "loss": 0.2309,
+      "step": 4914
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.8547551789077214e-05,
+      "loss": 0.1764,
+      "step": 4915
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.8545197740113e-05,
+      "loss": 0.1684,
+      "step": 4916
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.8542843691148775e-05,
+      "loss": 0.0892,
+      "step": 4917
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.854048964218456e-05,
+      "loss": 0.2117,
+      "step": 4918
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.853813559322034e-05,
+      "loss": 0.2686,
+      "step": 4919
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.853578154425612e-05,
+      "loss": 0.1655,
+      "step": 4920
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.8533427495291904e-05,
+      "loss": 0.2038,
+      "step": 4921
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.853107344632769e-05,
+      "loss": 0.2372,
+      "step": 4922
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.8528719397363465e-05,
+      "loss": 0.1218,
+      "step": 4923
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.852636534839925e-05,
+      "loss": 0.1917,
+      "step": 4924
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.852401129943503e-05,
+      "loss": 0.2022,
+      "step": 4925
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.852165725047081e-05,
+      "loss": 0.2398,
+      "step": 4926
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.8519303201506594e-05,
+      "loss": 0.268,
+      "step": 4927
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.851694915254238e-05,
+      "loss": 0.2019,
+      "step": 4928
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8514595103578155e-05,
+      "loss": 0.1784,
+      "step": 4929
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.851224105461394e-05,
+      "loss": 0.1421,
+      "step": 4930
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8509887005649716e-05,
+      "loss": 0.3212,
+      "step": 4931
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.85075329566855e-05,
+      "loss": 0.2186,
+      "step": 4932
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8505178907721284e-05,
+      "loss": 0.3141,
+      "step": 4933
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.850282485875706e-05,
+      "loss": 0.1008,
+      "step": 4934
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8500470809792845e-05,
+      "loss": 0.2328,
+      "step": 4935
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.849811676082863e-05,
+      "loss": 0.2028,
+      "step": 4936
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8495762711864406e-05,
+      "loss": 0.2693,
+      "step": 4937
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.849340866290019e-05,
+      "loss": 0.1364,
+      "step": 4938
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8491054613935974e-05,
+      "loss": 0.2154,
+      "step": 4939
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.848870056497175e-05,
+      "loss": 0.3,
+      "step": 4940
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8486346516007535e-05,
+      "loss": 0.1814,
+      "step": 4941
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.848399246704332e-05,
+      "loss": 0.1721,
+      "step": 4942
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8481638418079096e-05,
+      "loss": 0.1385,
+      "step": 4943
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.847928436911488e-05,
+      "loss": 0.2893,
+      "step": 4944
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8476930320150664e-05,
+      "loss": 0.2869,
+      "step": 4945
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.847457627118644e-05,
+      "loss": 0.1932,
+      "step": 4946
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8472222222222225e-05,
+      "loss": 0.2173,
+      "step": 4947
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8469868173258e-05,
+      "loss": 0.2245,
+      "step": 4948
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8467514124293786e-05,
+      "loss": 0.3559,
+      "step": 4949
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.846516007532957e-05,
+      "loss": 0.2809,
+      "step": 4950
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.846280602636535e-05,
+      "loss": 0.1679,
+      "step": 4951
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.846045197740113e-05,
+      "loss": 0.2106,
+      "step": 4952
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8458097928436915e-05,
+      "loss": 0.2144,
+      "step": 4953
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.845574387947269e-05,
+      "loss": 0.2743,
+      "step": 4954
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8453389830508476e-05,
+      "loss": 0.2602,
+      "step": 4955
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.845103578154426e-05,
+      "loss": 0.2914,
+      "step": 4956
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.844868173258004e-05,
+      "loss": 0.154,
+      "step": 4957
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.844632768361582e-05,
+      "loss": 0.2549,
+      "step": 4958
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8443973634651605e-05,
+      "loss": 0.2489,
+      "step": 4959
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.844161958568738e-05,
+      "loss": 0.2282,
+      "step": 4960
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8439265536723166e-05,
+      "loss": 0.2318,
+      "step": 4961
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.843691148775895e-05,
+      "loss": 0.2357,
+      "step": 4962
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.843455743879473e-05,
+      "loss": 0.1577,
+      "step": 4963
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.843220338983051e-05,
+      "loss": 0.2069,
+      "step": 4964
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8429849340866295e-05,
+      "loss": 0.2149,
+      "step": 4965
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.842749529190207e-05,
+      "loss": 0.1649,
+      "step": 4966
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8425141242937856e-05,
+      "loss": 0.175,
+      "step": 4967
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8422787193973633e-05,
+      "loss": 0.2977,
+      "step": 4968
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.842043314500942e-05,
+      "loss": 0.1552,
+      "step": 4969
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.84180790960452e-05,
+      "loss": 0.1427,
+      "step": 4970
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.841572504708098e-05,
+      "loss": 0.1877,
+      "step": 4971
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.841337099811676e-05,
+      "loss": 0.2319,
+      "step": 4972
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8411016949152546e-05,
+      "loss": 0.196,
+      "step": 4973
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8408662900188323e-05,
+      "loss": 0.3393,
+      "step": 4974
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.840630885122411e-05,
+      "loss": 0.3253,
+      "step": 4975
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.840395480225989e-05,
+      "loss": 0.0907,
+      "step": 4976
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.840160075329567e-05,
+      "loss": 0.1915,
+      "step": 4977
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.839924670433145e-05,
+      "loss": 0.1769,
+      "step": 4978
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8396892655367236e-05,
+      "loss": 0.2796,
+      "step": 4979
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8394538606403013e-05,
+      "loss": 0.0894,
+      "step": 4980
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.83921845574388e-05,
+      "loss": 0.1263,
+      "step": 4981
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.838983050847458e-05,
+      "loss": 0.1022,
+      "step": 4982
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.838747645951036e-05,
+      "loss": 0.2506,
+      "step": 4983
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.838512241054614e-05,
+      "loss": 0.2169,
+      "step": 4984
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.838276836158192e-05,
+      "loss": 0.1216,
+      "step": 4985
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8380414312617703e-05,
+      "loss": 0.2563,
+      "step": 4986
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.837806026365349e-05,
+      "loss": 0.152,
+      "step": 4987
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8375706214689265e-05,
+      "loss": 0.2359,
+      "step": 4988
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.837335216572505e-05,
+      "loss": 0.2177,
+      "step": 4989
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.837099811676083e-05,
+      "loss": 0.1428,
+      "step": 4990
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.836864406779661e-05,
+      "loss": 0.0822,
+      "step": 4991
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8366290018832393e-05,
+      "loss": 0.299,
+      "step": 4992
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.836393596986818e-05,
+      "loss": 0.1852,
+      "step": 4993
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.8361581920903955e-05,
+      "loss": 0.1671,
+      "step": 4994
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.835922787193974e-05,
+      "loss": 0.2051,
+      "step": 4995
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.835687382297552e-05,
+      "loss": 0.2312,
+      "step": 4996
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.83545197740113e-05,
+      "loss": 0.2937,
+      "step": 4997
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.8352165725047083e-05,
+      "loss": 0.123,
+      "step": 4998
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.834981167608287e-05,
+      "loss": 0.1262,
+      "step": 4999
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.8347457627118644e-05,
+      "loss": 0.2162,
+      "step": 5000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.834510357815443e-05,
+      "loss": 0.1535,
+      "step": 5001
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.8342749529190206e-05,
+      "loss": 0.0551,
+      "step": 5002
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.834039548022599e-05,
+      "loss": 0.4024,
+      "step": 5003
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.8338041431261773e-05,
+      "loss": 0.3053,
+      "step": 5004
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.833568738229755e-05,
+      "loss": 0.4438,
+      "step": 5005
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.8333333333333334e-05,
+      "loss": 0.2794,
+      "step": 5006
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.833097928436912e-05,
+      "loss": 0.2192,
+      "step": 5007
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.8328625235404896e-05,
+      "loss": 0.2455,
+      "step": 5008
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.832627118644068e-05,
+      "loss": 0.3132,
+      "step": 5009
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.8323917137476463e-05,
+      "loss": 0.2463,
+      "step": 5010
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.832156308851224e-05,
+      "loss": 0.2746,
+      "step": 5011
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.8319209039548024e-05,
+      "loss": 0.3083,
+      "step": 5012
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.831685499058381e-05,
+      "loss": 0.3185,
+      "step": 5013
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8314500941619586e-05,
+      "loss": 0.2238,
+      "step": 5014
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.831214689265537e-05,
+      "loss": 0.2342,
+      "step": 5015
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8309792843691153e-05,
+      "loss": 0.3076,
+      "step": 5016
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.830743879472693e-05,
+      "loss": 0.1478,
+      "step": 5017
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8305084745762714e-05,
+      "loss": 0.3352,
+      "step": 5018
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.830273069679849e-05,
+      "loss": 0.1807,
+      "step": 5019
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8300376647834276e-05,
+      "loss": 0.1412,
+      "step": 5020
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.829802259887006e-05,
+      "loss": 0.1132,
+      "step": 5021
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.829566854990584e-05,
+      "loss": 0.1897,
+      "step": 5022
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.829331450094162e-05,
+      "loss": 0.1714,
+      "step": 5023
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8290960451977404e-05,
+      "loss": 0.1267,
+      "step": 5024
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.828860640301318e-05,
+      "loss": 0.2651,
+      "step": 5025
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8286252354048966e-05,
+      "loss": 0.2917,
+      "step": 5026
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.828389830508475e-05,
+      "loss": 0.2148,
+      "step": 5027
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.828154425612053e-05,
+      "loss": 0.1937,
+      "step": 5028
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.827919020715631e-05,
+      "loss": 0.1726,
+      "step": 5029
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8276836158192094e-05,
+      "loss": 0.147,
+      "step": 5030
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.827448210922787e-05,
+      "loss": 0.1545,
+      "step": 5031
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8272128060263656e-05,
+      "loss": 0.179,
+      "step": 5032
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.826977401129944e-05,
+      "loss": 0.1284,
+      "step": 5033
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.826741996233522e-05,
+      "loss": 0.237,
+      "step": 5034
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8265065913371e-05,
+      "loss": 0.1967,
+      "step": 5035
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.826271186440678e-05,
+      "loss": 0.3083,
+      "step": 5036
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.826035781544257e-05,
+      "loss": 0.1328,
+      "step": 5037
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8258003766478346e-05,
+      "loss": 0.2701,
+      "step": 5038
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.825564971751412e-05,
+      "loss": 0.1252,
+      "step": 5039
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8253295668549907e-05,
+      "loss": 0.1573,
+      "step": 5040
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.825094161958569e-05,
+      "loss": 0.2872,
+      "step": 5041
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.824858757062147e-05,
+      "loss": 0.1865,
+      "step": 5042
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.824623352165725e-05,
+      "loss": 0.14,
+      "step": 5043
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8243879472693036e-05,
+      "loss": 0.1,
+      "step": 5044
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.824152542372881e-05,
+      "loss": 0.2713,
+      "step": 5045
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8239171374764597e-05,
+      "loss": 0.2721,
+      "step": 5046
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.823681732580038e-05,
+      "loss": 0.2559,
+      "step": 5047
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.823446327683616e-05,
+      "loss": 0.2893,
+      "step": 5048
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.823210922787194e-05,
+      "loss": 0.187,
+      "step": 5049
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8229755178907726e-05,
+      "loss": 0.1618,
+      "step": 5050
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.82274011299435e-05,
+      "loss": 0.1275,
+      "step": 5051
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8225047080979287e-05,
+      "loss": 0.2402,
+      "step": 5052
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8222693032015064e-05,
+      "loss": 0.1985,
+      "step": 5053
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8220338983050854e-05,
+      "loss": 0.3702,
+      "step": 5054
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.821798493408663e-05,
+      "loss": 0.1696,
+      "step": 5055
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.821563088512241e-05,
+      "loss": 0.1649,
+      "step": 5056
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.82132768361582e-05,
+      "loss": 0.2398,
+      "step": 5057
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8210922787193977e-05,
+      "loss": 0.2567,
+      "step": 5058
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8208568738229754e-05,
+      "loss": 0.0856,
+      "step": 5059
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.820621468926554e-05,
+      "loss": 0.2828,
+      "step": 5060
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.820386064030132e-05,
+      "loss": 0.1406,
+      "step": 5061
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.82015065913371e-05,
+      "loss": 0.2675,
+      "step": 5062
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.819915254237288e-05,
+      "loss": 0.1789,
+      "step": 5063
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8196798493408667e-05,
+      "loss": 0.1615,
+      "step": 5064
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8194444444444444e-05,
+      "loss": 0.1757,
+      "step": 5065
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.819209039548023e-05,
+      "loss": 0.208,
+      "step": 5066
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.818973634651601e-05,
+      "loss": 0.2263,
+      "step": 5067
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8187382297551796e-05,
+      "loss": 0.0778,
+      "step": 5068
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.818502824858757e-05,
+      "loss": 0.3159,
+      "step": 5069
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.818267419962335e-05,
+      "loss": 0.1849,
+      "step": 5070
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.818032015065914e-05,
+      "loss": 0.2587,
+      "step": 5071
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.817796610169492e-05,
+      "loss": 0.3305,
+      "step": 5072
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8175612052730695e-05,
+      "loss": 0.1335,
+      "step": 5073
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8173258003766486e-05,
+      "loss": 0.1607,
+      "step": 5074
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.817090395480226e-05,
+      "loss": 0.1304,
+      "step": 5075
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.816854990583804e-05,
+      "loss": 0.2503,
+      "step": 5076
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8166195856873824e-05,
+      "loss": 0.2568,
+      "step": 5077
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.816384180790961e-05,
+      "loss": 0.2825,
+      "step": 5078
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8161487758945385e-05,
+      "loss": 0.2548,
+      "step": 5079
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.815913370998117e-05,
+      "loss": 0.2183,
+      "step": 5080
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.815677966101695e-05,
+      "loss": 0.1883,
+      "step": 5081
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.815442561205273e-05,
+      "loss": 0.1677,
+      "step": 5082
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8152071563088514e-05,
+      "loss": 0.2491,
+      "step": 5083
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.81497175141243e-05,
+      "loss": 0.2412,
+      "step": 5084
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.814736346516008e-05,
+      "loss": 0.2029,
+      "step": 5085
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.814500941619586e-05,
+      "loss": 0.2057,
+      "step": 5086
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8142655367231636e-05,
+      "loss": 0.2282,
+      "step": 5087
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8140301318267427e-05,
+      "loss": 0.2387,
+      "step": 5088
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8137947269303204e-05,
+      "loss": 0.3276,
+      "step": 5089
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.813559322033898e-05,
+      "loss": 0.1593,
+      "step": 5090
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.813323917137477e-05,
+      "loss": 0.2424,
+      "step": 5091
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.813088512241055e-05,
+      "loss": 0.2155,
+      "step": 5092
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8128531073446326e-05,
+      "loss": 0.3207,
+      "step": 5093
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.812617702448211e-05,
+      "loss": 0.3492,
+      "step": 5094
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8123822975517894e-05,
+      "loss": 0.2034,
+      "step": 5095
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.812146892655367e-05,
+      "loss": 0.3132,
+      "step": 5096
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8119114877589455e-05,
+      "loss": 0.2453,
+      "step": 5097
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.811676082862524e-05,
+      "loss": 0.1957,
+      "step": 5098
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8114406779661016e-05,
+      "loss": 0.1738,
+      "step": 5099
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.81120527306968e-05,
+      "loss": 0.1472,
+      "step": 5100
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8109698681732584e-05,
+      "loss": 0.2677,
+      "step": 5101
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.810734463276837e-05,
+      "loss": 0.2277,
+      "step": 5102
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8104990583804145e-05,
+      "loss": 0.2017,
+      "step": 5103
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.810263653483992e-05,
+      "loss": 0.1866,
+      "step": 5104
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.810028248587571e-05,
+      "loss": 0.2112,
+      "step": 5105
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.809792843691149e-05,
+      "loss": 0.3204,
+      "step": 5106
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.809557438794727e-05,
+      "loss": 0.137,
+      "step": 5107
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.809322033898306e-05,
+      "loss": 0.3702,
+      "step": 5108
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8090866290018835e-05,
+      "loss": 0.1461,
+      "step": 5109
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.808851224105461e-05,
+      "loss": 0.2755,
+      "step": 5110
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8086158192090396e-05,
+      "loss": 0.1218,
+      "step": 5111
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.808380414312618e-05,
+      "loss": 0.231,
+      "step": 5112
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.808145009416196e-05,
+      "loss": 0.2516,
+      "step": 5113
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.807909604519774e-05,
+      "loss": 0.2475,
+      "step": 5114
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8076741996233525e-05,
+      "loss": 0.3229,
+      "step": 5115
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.807438794726931e-05,
+      "loss": 0.2689,
+      "step": 5116
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8072033898305086e-05,
+      "loss": 0.2529,
+      "step": 5117
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.806967984934087e-05,
+      "loss": 0.2834,
+      "step": 5118
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8067325800376654e-05,
+      "loss": 0.162,
+      "step": 5119
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.806497175141243e-05,
+      "loss": 0.2664,
+      "step": 5120
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.806261770244821e-05,
+      "loss": 0.275,
+      "step": 5121
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.8060263653484e-05,
+      "loss": 0.1664,
+      "step": 5122
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.8057909604519776e-05,
+      "loss": 0.3153,
+      "step": 5123
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.805555555555555e-05,
+      "loss": 0.2531,
+      "step": 5124
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.8053201506591344e-05,
+      "loss": 0.2194,
+      "step": 5125
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.805084745762712e-05,
+      "loss": 0.1999,
+      "step": 5126
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.80484934086629e-05,
+      "loss": 0.1812,
+      "step": 5127
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.804613935969868e-05,
+      "loss": 0.2153,
+      "step": 5128
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.8043785310734466e-05,
+      "loss": 0.1492,
+      "step": 5129
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.804143126177024e-05,
+      "loss": 0.1175,
+      "step": 5130
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.803907721280603e-05,
+      "loss": 0.2733,
+      "step": 5131
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.803672316384181e-05,
+      "loss": 0.2011,
+      "step": 5132
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.8034369114877595e-05,
+      "loss": 0.2821,
+      "step": 5133
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.803201506591337e-05,
+      "loss": 0.3749,
+      "step": 5134
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.8029661016949156e-05,
+      "loss": 0.1683,
+      "step": 5135
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.802730696798494e-05,
+      "loss": 0.219,
+      "step": 5136
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.802495291902072e-05,
+      "loss": 0.3471,
+      "step": 5137
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.8022598870056494e-05,
+      "loss": 0.2288,
+      "step": 5138
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.8020244821092285e-05,
+      "loss": 0.2211,
+      "step": 5139
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.801789077212806e-05,
+      "loss": 0.2428,
+      "step": 5140
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.801553672316384e-05,
+      "loss": 0.1383,
+      "step": 5141
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.801318267419963e-05,
+      "loss": 0.2007,
+      "step": 5142
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.801082862523541e-05,
+      "loss": 0.4164,
+      "step": 5143
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8008474576271184e-05,
+      "loss": 0.1911,
+      "step": 5144
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.800612052730697e-05,
+      "loss": 0.196,
+      "step": 5145
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.800376647834275e-05,
+      "loss": 0.2157,
+      "step": 5146
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8001412429378536e-05,
+      "loss": 0.1841,
+      "step": 5147
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.799905838041431e-05,
+      "loss": 0.1801,
+      "step": 5148
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.79967043314501e-05,
+      "loss": 0.196,
+      "step": 5149
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.799435028248588e-05,
+      "loss": 0.254,
+      "step": 5150
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.799199623352166e-05,
+      "loss": 0.1529,
+      "step": 5151
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.798964218455744e-05,
+      "loss": 0.1451,
+      "step": 5152
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7987288135593226e-05,
+      "loss": 0.136,
+      "step": 5153
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7984934086629e-05,
+      "loss": 0.2038,
+      "step": 5154
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.798258003766478e-05,
+      "loss": 0.2671,
+      "step": 5155
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.798022598870057e-05,
+      "loss": 0.2803,
+      "step": 5156
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.797787193973635e-05,
+      "loss": 0.2166,
+      "step": 5157
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7975517890772125e-05,
+      "loss": 0.1314,
+      "step": 5158
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7973163841807916e-05,
+      "loss": 0.2411,
+      "step": 5159
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.797080979284369e-05,
+      "loss": 0.3417,
+      "step": 5160
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.796845574387947e-05,
+      "loss": 0.3112,
+      "step": 5161
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7966101694915254e-05,
+      "loss": 0.3478,
+      "step": 5162
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.796374764595104e-05,
+      "loss": 0.1933,
+      "step": 5163
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.796139359698682e-05,
+      "loss": 0.1619,
+      "step": 5164
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.79590395480226e-05,
+      "loss": 0.2465,
+      "step": 5165
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.795668549905838e-05,
+      "loss": 0.1744,
+      "step": 5166
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.795433145009417e-05,
+      "loss": 0.2106,
+      "step": 5167
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7951977401129944e-05,
+      "loss": 0.242,
+      "step": 5168
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.794962335216573e-05,
+      "loss": 0.1786,
+      "step": 5169
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.794726930320151e-05,
+      "loss": 0.2407,
+      "step": 5170
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.794491525423729e-05,
+      "loss": 0.3423,
+      "step": 5171
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7942561205273066e-05,
+      "loss": 0.3609,
+      "step": 5172
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.794020715630886e-05,
+      "loss": 0.1499,
+      "step": 5173
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7937853107344634e-05,
+      "loss": 0.2138,
+      "step": 5174
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.793549905838041e-05,
+      "loss": 0.2558,
+      "step": 5175
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.79331450094162e-05,
+      "loss": 0.2308,
+      "step": 5176
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.793079096045198e-05,
+      "loss": 0.291,
+      "step": 5177
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7928436911487756e-05,
+      "loss": 0.1769,
+      "step": 5178
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.792608286252354e-05,
+      "loss": 0.2709,
+      "step": 5179
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7923728813559324e-05,
+      "loss": 0.1834,
+      "step": 5180
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.792137476459511e-05,
+      "loss": 0.1584,
+      "step": 5181
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7919020715630885e-05,
+      "loss": 0.3099,
+      "step": 5182
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.791666666666667e-05,
+      "loss": 0.2992,
+      "step": 5183
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.791431261770245e-05,
+      "loss": 0.2988,
+      "step": 5184
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.791195856873823e-05,
+      "loss": 0.1744,
+      "step": 5185
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7909604519774014e-05,
+      "loss": 0.2561,
+      "step": 5186
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.79072504708098e-05,
+      "loss": 0.1336,
+      "step": 5187
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7904896421845575e-05,
+      "loss": 0.2252,
+      "step": 5188
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.790254237288135e-05,
+      "loss": 0.1958,
+      "step": 5189
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.790018832391714e-05,
+      "loss": 0.1974,
+      "step": 5190
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.789783427495292e-05,
+      "loss": 0.1468,
+      "step": 5191
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.78954802259887e-05,
+      "loss": 0.2044,
+      "step": 5192
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.789312617702449e-05,
+      "loss": 0.1546,
+      "step": 5193
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7890772128060265e-05,
+      "loss": 0.2174,
+      "step": 5194
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.788841807909605e-05,
+      "loss": 0.2188,
+      "step": 5195
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7886064030131826e-05,
+      "loss": 0.2443,
+      "step": 5196
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.788370998116761e-05,
+      "loss": 0.2101,
+      "step": 5197
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7881355932203394e-05,
+      "loss": 0.1561,
+      "step": 5198
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.787900188323917e-05,
+      "loss": 0.1743,
+      "step": 5199
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7876647834274955e-05,
+      "loss": 0.2213,
+      "step": 5200
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.787429378531074e-05,
+      "loss": 0.2309,
+      "step": 5201
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7871939736346516e-05,
+      "loss": 0.2644,
+      "step": 5202
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.78695856873823e-05,
+      "loss": 0.2586,
+      "step": 5203
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7867231638418084e-05,
+      "loss": 0.1213,
+      "step": 5204
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.786487758945386e-05,
+      "loss": 0.305,
+      "step": 5205
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7862523540489645e-05,
+      "loss": 0.2639,
+      "step": 5206
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.786016949152543e-05,
+      "loss": 0.2567,
+      "step": 5207
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7857815442561206e-05,
+      "loss": 0.1568,
+      "step": 5208
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.785546139359698e-05,
+      "loss": 0.2352,
+      "step": 5209
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7853107344632774e-05,
+      "loss": 0.1479,
+      "step": 5210
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.785075329566855e-05,
+      "loss": 0.2778,
+      "step": 5211
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7848399246704335e-05,
+      "loss": 0.2076,
+      "step": 5212
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.784604519774011e-05,
+      "loss": 0.2412,
+      "step": 5213
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7843691148775896e-05,
+      "loss": 0.2917,
+      "step": 5214
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.784133709981168e-05,
+      "loss": 0.5196,
+      "step": 5215
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.783898305084746e-05,
+      "loss": 0.1311,
+      "step": 5216
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.783662900188324e-05,
+      "loss": 0.3063,
+      "step": 5217
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7834274952919025e-05,
+      "loss": 0.1534,
+      "step": 5218
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.78319209039548e-05,
+      "loss": 0.2461,
+      "step": 5219
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7829566854990586e-05,
+      "loss": 0.2185,
+      "step": 5220
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.782721280602637e-05,
+      "loss": 0.1843,
+      "step": 5221
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.782485875706215e-05,
+      "loss": 0.268,
+      "step": 5222
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.782250470809793e-05,
+      "loss": 0.1418,
+      "step": 5223
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7820150659133715e-05,
+      "loss": 0.1478,
+      "step": 5224
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.781779661016949e-05,
+      "loss": 0.2125,
+      "step": 5225
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7815442561205276e-05,
+      "loss": 0.1783,
+      "step": 5226
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.781308851224106e-05,
+      "loss": 0.1353,
+      "step": 5227
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.781073446327684e-05,
+      "loss": 0.1494,
+      "step": 5228
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.780838041431262e-05,
+      "loss": 0.1643,
+      "step": 5229
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.78060263653484e-05,
+      "loss": 0.1442,
+      "step": 5230
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.780367231638418e-05,
+      "loss": 0.3039,
+      "step": 5231
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7801318267419966e-05,
+      "loss": 0.2014,
+      "step": 5232
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.779896421845574e-05,
+      "loss": 0.4265,
+      "step": 5233
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.779661016949153e-05,
+      "loss": 0.1482,
+      "step": 5234
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.779425612052731e-05,
+      "loss": 0.145,
+      "step": 5235
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.779190207156309e-05,
+      "loss": 0.261,
+      "step": 5236
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.778954802259887e-05,
+      "loss": 0.1587,
+      "step": 5237
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7787193973634656e-05,
+      "loss": 0.1615,
+      "step": 5238
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.778483992467043e-05,
+      "loss": 0.249,
+      "step": 5239
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.778248587570622e-05,
+      "loss": 0.1543,
+      "step": 5240
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7780131826742e-05,
+      "loss": 0.1765,
+      "step": 5241
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.777777777777778e-05,
+      "loss": 0.1426,
+      "step": 5242
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.777542372881356e-05,
+      "loss": 0.1719,
+      "step": 5243
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7773069679849346e-05,
+      "loss": 0.268,
+      "step": 5244
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.777071563088512e-05,
+      "loss": 0.2973,
+      "step": 5245
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.776836158192091e-05,
+      "loss": 0.1394,
+      "step": 5246
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7766007532956684e-05,
+      "loss": 0.3255,
+      "step": 5247
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.776365348399247e-05,
+      "loss": 0.1267,
+      "step": 5248
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.776129943502825e-05,
+      "loss": 0.2264,
+      "step": 5249
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.775894538606403e-05,
+      "loss": 0.2083,
+      "step": 5250
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.775659133709981e-05,
+      "loss": 0.208,
+      "step": 5251
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.77542372881356e-05,
+      "loss": 0.2463,
+      "step": 5252
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7751883239171374e-05,
+      "loss": 0.2215,
+      "step": 5253
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.774952919020716e-05,
+      "loss": 0.318,
+      "step": 5254
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.774717514124294e-05,
+      "loss": 0.2009,
+      "step": 5255
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.774482109227872e-05,
+      "loss": 0.1791,
+      "step": 5256
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.77424670433145e-05,
+      "loss": 0.1276,
+      "step": 5257
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.774011299435029e-05,
+      "loss": 0.2044,
+      "step": 5258
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7737758945386064e-05,
+      "loss": 0.1443,
+      "step": 5259
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.773540489642185e-05,
+      "loss": 0.1727,
+      "step": 5260
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.773305084745763e-05,
+      "loss": 0.1516,
+      "step": 5261
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.773069679849341e-05,
+      "loss": 0.1934,
+      "step": 5262
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.772834274952919e-05,
+      "loss": 0.3352,
+      "step": 5263
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.772598870056497e-05,
+      "loss": 0.1586,
+      "step": 5264
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7723634651600754e-05,
+      "loss": 0.3079,
+      "step": 5265
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.772128060263654e-05,
+      "loss": 0.0651,
+      "step": 5266
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7718926553672315e-05,
+      "loss": 0.2673,
+      "step": 5267
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.77165725047081e-05,
+      "loss": 0.1548,
+      "step": 5268
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.771421845574388e-05,
+      "loss": 0.1431,
+      "step": 5269
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.771186440677966e-05,
+      "loss": 0.2832,
+      "step": 5270
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7709510357815444e-05,
+      "loss": 0.1872,
+      "step": 5271
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.770715630885123e-05,
+      "loss": 0.1654,
+      "step": 5272
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7704802259887005e-05,
+      "loss": 0.1877,
+      "step": 5273
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.770244821092279e-05,
+      "loss": 0.1726,
+      "step": 5274
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.770009416195857e-05,
+      "loss": 0.3204,
+      "step": 5275
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.769774011299435e-05,
+      "loss": 0.239,
+      "step": 5276
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7695386064030134e-05,
+      "loss": 0.2466,
+      "step": 5277
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.769303201506592e-05,
+      "loss": 0.2898,
+      "step": 5278
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7690677966101695e-05,
+      "loss": 0.249,
+      "step": 5279
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.768832391713748e-05,
+      "loss": 0.2102,
+      "step": 5280
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.768596986817326e-05,
+      "loss": 0.2883,
+      "step": 5281
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.768361581920904e-05,
+      "loss": 0.3213,
+      "step": 5282
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7681261770244824e-05,
+      "loss": 0.1687,
+      "step": 5283
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.76789077212806e-05,
+      "loss": 0.131,
+      "step": 5284
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7676553672316385e-05,
+      "loss": 0.1202,
+      "step": 5285
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.767419962335217e-05,
+      "loss": 0.1676,
+      "step": 5286
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7671845574387946e-05,
+      "loss": 0.2921,
+      "step": 5287
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.766949152542373e-05,
+      "loss": 0.1435,
+      "step": 5288
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7667137476459514e-05,
+      "loss": 0.1266,
+      "step": 5289
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.766478342749529e-05,
+      "loss": 0.1269,
+      "step": 5290
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7662429378531075e-05,
+      "loss": 0.3334,
+      "step": 5291
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.766007532956686e-05,
+      "loss": 0.2546,
+      "step": 5292
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7657721280602636e-05,
+      "loss": 0.205,
+      "step": 5293
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.765536723163842e-05,
+      "loss": 0.1811,
+      "step": 5294
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7653013182674204e-05,
+      "loss": 0.1149,
+      "step": 5295
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.765065913370998e-05,
+      "loss": 0.2415,
+      "step": 5296
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7648305084745765e-05,
+      "loss": 0.1041,
+      "step": 5297
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.764595103578155e-05,
+      "loss": 0.2338,
+      "step": 5298
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7643596986817326e-05,
+      "loss": 0.0982,
+      "step": 5299
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.764124293785311e-05,
+      "loss": 0.245,
+      "step": 5300
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.763888888888889e-05,
+      "loss": 0.2447,
+      "step": 5301
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.763653483992467e-05,
+      "loss": 0.126,
+      "step": 5302
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7634180790960455e-05,
+      "loss": 0.1625,
+      "step": 5303
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.763182674199623e-05,
+      "loss": 0.1531,
+      "step": 5304
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7629472693032016e-05,
+      "loss": 0.133,
+      "step": 5305
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.76271186440678e-05,
+      "loss": 0.1976,
+      "step": 5306
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.762476459510358e-05,
+      "loss": 0.2504,
+      "step": 5307
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.762241054613936e-05,
+      "loss": 0.0985,
+      "step": 5308
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7620056497175145e-05,
+      "loss": 0.2123,
+      "step": 5309
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.761770244821092e-05,
+      "loss": 0.3534,
+      "step": 5310
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7615348399246706e-05,
+      "loss": 0.081,
+      "step": 5311
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.761299435028249e-05,
+      "loss": 0.2216,
+      "step": 5312
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.761064030131827e-05,
+      "loss": 0.111,
+      "step": 5313
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.760828625235405e-05,
+      "loss": 0.2328,
+      "step": 5314
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7605932203389835e-05,
+      "loss": 0.202,
+      "step": 5315
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.760357815442561e-05,
+      "loss": 0.2368,
+      "step": 5316
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7601224105461396e-05,
+      "loss": 0.2707,
+      "step": 5317
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7598870056497174e-05,
+      "loss": 0.2702,
+      "step": 5318
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.759651600753296e-05,
+      "loss": 0.1566,
+      "step": 5319
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.759416195856874e-05,
+      "loss": 0.287,
+      "step": 5320
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.759180790960452e-05,
+      "loss": 0.1548,
+      "step": 5321
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.75894538606403e-05,
+      "loss": 0.1873,
+      "step": 5322
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7587099811676086e-05,
+      "loss": 0.1341,
+      "step": 5323
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7584745762711864e-05,
+      "loss": 0.2377,
+      "step": 5324
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.758239171374765e-05,
+      "loss": 0.1375,
+      "step": 5325
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.758003766478343e-05,
+      "loss": 0.3764,
+      "step": 5326
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.757768361581921e-05,
+      "loss": 0.3187,
+      "step": 5327
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.757532956685499e-05,
+      "loss": 0.2996,
+      "step": 5328
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7572975517890776e-05,
+      "loss": 0.1524,
+      "step": 5329
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7570621468926554e-05,
+      "loss": 0.1808,
+      "step": 5330
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.756826741996234e-05,
+      "loss": 0.2664,
+      "step": 5331
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.756591337099812e-05,
+      "loss": 0.2833,
+      "step": 5332
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.75635593220339e-05,
+      "loss": 0.4558,
+      "step": 5333
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.756120527306968e-05,
+      "loss": 0.245,
+      "step": 5334
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.755885122410546e-05,
+      "loss": 0.23,
+      "step": 5335
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7556497175141244e-05,
+      "loss": 0.1684,
+      "step": 5336
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.755414312617703e-05,
+      "loss": 0.1377,
+      "step": 5337
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7551789077212805e-05,
+      "loss": 0.1182,
+      "step": 5338
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.754943502824859e-05,
+      "loss": 0.1925,
+      "step": 5339
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.754708097928437e-05,
+      "loss": 0.1586,
+      "step": 5340
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.754472693032015e-05,
+      "loss": 0.1479,
+      "step": 5341
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7542372881355934e-05,
+      "loss": 0.2101,
+      "step": 5342
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.754001883239172e-05,
+      "loss": 0.2541,
+      "step": 5343
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7537664783427495e-05,
+      "loss": 0.163,
+      "step": 5344
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.753531073446328e-05,
+      "loss": 0.1791,
+      "step": 5345
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.753295668549906e-05,
+      "loss": 0.1707,
+      "step": 5346
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.753060263653484e-05,
+      "loss": 0.2405,
+      "step": 5347
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7528248587570624e-05,
+      "loss": 0.3055,
+      "step": 5348
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.752589453860641e-05,
+      "loss": 0.173,
+      "step": 5349
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7523540489642185e-05,
+      "loss": 0.126,
+      "step": 5350
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.752118644067797e-05,
+      "loss": 0.2205,
+      "step": 5351
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7518832391713746e-05,
+      "loss": 0.1547,
+      "step": 5352
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7516478342749536e-05,
+      "loss": 0.2907,
+      "step": 5353
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7514124293785313e-05,
+      "loss": 0.2889,
+      "step": 5354
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.751177024482109e-05,
+      "loss": 0.1201,
+      "step": 5355
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7509416195856875e-05,
+      "loss": 0.1445,
+      "step": 5356
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.750706214689266e-05,
+      "loss": 0.229,
+      "step": 5357
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7504708097928436e-05,
+      "loss": 0.1209,
+      "step": 5358
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.750235404896422e-05,
+      "loss": 0.3016,
+      "step": 5359
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.1082,
+      "step": 5360
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.749764595103578e-05,
+      "loss": 0.125,
+      "step": 5361
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7495291902071565e-05,
+      "loss": 0.1928,
+      "step": 5362
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.749293785310735e-05,
+      "loss": 0.2935,
+      "step": 5363
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7490583804143126e-05,
+      "loss": 0.2577,
+      "step": 5364
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.748822975517891e-05,
+      "loss": 0.2068,
+      "step": 5365
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7485875706214693e-05,
+      "loss": 0.2629,
+      "step": 5366
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.748352165725047e-05,
+      "loss": 0.2853,
+      "step": 5367
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7481167608286255e-05,
+      "loss": 0.2393,
+      "step": 5368
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.747881355932203e-05,
+      "loss": 0.1378,
+      "step": 5369
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.747645951035782e-05,
+      "loss": 0.2506,
+      "step": 5370
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.74741054613936e-05,
+      "loss": 0.2325,
+      "step": 5371
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.747175141242938e-05,
+      "loss": 0.1709,
+      "step": 5372
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.746939736346517e-05,
+      "loss": 0.288,
+      "step": 5373
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7467043314500945e-05,
+      "loss": 0.2095,
+      "step": 5374
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.746468926553672e-05,
+      "loss": 0.1095,
+      "step": 5375
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7462335216572506e-05,
+      "loss": 0.1699,
+      "step": 5376
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.745998116760829e-05,
+      "loss": 0.1269,
+      "step": 5377
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.745762711864407e-05,
+      "loss": 0.2693,
+      "step": 5378
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.745527306967985e-05,
+      "loss": 0.2339,
+      "step": 5379
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7452919020715635e-05,
+      "loss": 0.1031,
+      "step": 5380
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.745056497175141e-05,
+      "loss": 0.1191,
+      "step": 5381
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7448210922787196e-05,
+      "loss": 0.2502,
+      "step": 5382
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.744585687382298e-05,
+      "loss": 0.2027,
+      "step": 5383
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7443502824858763e-05,
+      "loss": 0.0801,
+      "step": 5384
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.744114877589454e-05,
+      "loss": 0.1,
+      "step": 5385
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.743879472693032e-05,
+      "loss": 0.318,
+      "step": 5386
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.743644067796611e-05,
+      "loss": 0.1625,
+      "step": 5387
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7434086629001886e-05,
+      "loss": 0.2986,
+      "step": 5388
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.743173258003766e-05,
+      "loss": 0.2541,
+      "step": 5389
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7429378531073453e-05,
+      "loss": 0.1289,
+      "step": 5390
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.742702448210923e-05,
+      "loss": 0.1872,
+      "step": 5391
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.742467043314501e-05,
+      "loss": 0.2155,
+      "step": 5392
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.742231638418079e-05,
+      "loss": 0.2073,
+      "step": 5393
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7419962335216576e-05,
+      "loss": 0.2027,
+      "step": 5394
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.741760828625235e-05,
+      "loss": 0.21,
+      "step": 5395
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.741525423728814e-05,
+      "loss": 0.1416,
+      "step": 5396
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.741290018832392e-05,
+      "loss": 0.2937,
+      "step": 5397
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.74105461393597e-05,
+      "loss": 0.1184,
+      "step": 5398
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.740819209039548e-05,
+      "loss": 0.1705,
+      "step": 5399
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7405838041431266e-05,
+      "loss": 0.0855,
+      "step": 5400
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.740348399246705e-05,
+      "loss": 0.1424,
+      "step": 5401
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.740112994350283e-05,
+      "loss": 0.3256,
+      "step": 5402
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7398775894538604e-05,
+      "loss": 0.0959,
+      "step": 5403
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7396421845574395e-05,
+      "loss": 0.1878,
+      "step": 5404
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.739406779661017e-05,
+      "loss": 0.0883,
+      "step": 5405
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.739171374764595e-05,
+      "loss": 0.1043,
+      "step": 5406
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.738935969868174e-05,
+      "loss": 0.1708,
+      "step": 5407
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.738700564971752e-05,
+      "loss": 0.1294,
+      "step": 5408
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7384651600753294e-05,
+      "loss": 0.2304,
+      "step": 5409
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.738229755178908e-05,
+      "loss": 0.2476,
+      "step": 5410
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.737994350282486e-05,
+      "loss": 0.1616,
+      "step": 5411
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.737758945386064e-05,
+      "loss": 0.2692,
+      "step": 5412
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.737523540489642e-05,
+      "loss": 0.3628,
+      "step": 5413
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.737288135593221e-05,
+      "loss": 0.2168,
+      "step": 5414
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7370527306967984e-05,
+      "loss": 0.2876,
+      "step": 5415
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.736817325800377e-05,
+      "loss": 0.2374,
+      "step": 5416
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.736581920903955e-05,
+      "loss": 0.2878,
+      "step": 5417
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7363465160075336e-05,
+      "loss": 0.2656,
+      "step": 5418
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.736111111111111e-05,
+      "loss": 0.2623,
+      "step": 5419
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.735875706214689e-05,
+      "loss": 0.2144,
+      "step": 5420
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.735640301318268e-05,
+      "loss": 0.3751,
+      "step": 5421
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.735404896421846e-05,
+      "loss": 0.3425,
+      "step": 5422
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7351694915254235e-05,
+      "loss": 0.0919,
+      "step": 5423
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7349340866290026e-05,
+      "loss": 0.1198,
+      "step": 5424
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.73469868173258e-05,
+      "loss": 0.2448,
+      "step": 5425
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.734463276836158e-05,
+      "loss": 0.2407,
+      "step": 5426
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7342278719397364e-05,
+      "loss": 0.2309,
+      "step": 5427
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.733992467043315e-05,
+      "loss": 0.1557,
+      "step": 5428
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7337570621468925e-05,
+      "loss": 0.1871,
+      "step": 5429
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.733521657250471e-05,
+      "loss": 0.1915,
+      "step": 5430
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.733286252354049e-05,
+      "loss": 0.2628,
+      "step": 5431
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.733050847457628e-05,
+      "loss": 0.1555,
+      "step": 5432
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7328154425612054e-05,
+      "loss": 0.1203,
+      "step": 5433
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.732580037664784e-05,
+      "loss": 0.2121,
+      "step": 5434
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.732344632768362e-05,
+      "loss": 0.3384,
+      "step": 5435
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.73210922787194e-05,
+      "loss": 0.2822,
+      "step": 5436
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7318738229755176e-05,
+      "loss": 0.1976,
+      "step": 5437
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.731638418079097e-05,
+      "loss": 0.2982,
+      "step": 5438
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7314030131826744e-05,
+      "loss": 0.1444,
+      "step": 5439
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.731167608286252e-05,
+      "loss": 0.1168,
+      "step": 5440
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.730932203389831e-05,
+      "loss": 0.2612,
+      "step": 5441
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.730696798493409e-05,
+      "loss": 0.0726,
+      "step": 5442
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7304613935969866e-05,
+      "loss": 0.3561,
+      "step": 5443
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.730225988700565e-05,
+      "loss": 0.3722,
+      "step": 5444
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7299905838041434e-05,
+      "loss": 0.2409,
+      "step": 5445
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.729755178907721e-05,
+      "loss": 0.1841,
+      "step": 5446
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7295197740112995e-05,
+      "loss": 0.1465,
+      "step": 5447
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.729284369114878e-05,
+      "loss": 0.0764,
+      "step": 5448
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.729048964218456e-05,
+      "loss": 0.2268,
+      "step": 5449
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.728813559322034e-05,
+      "loss": 0.3696,
+      "step": 5450
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7285781544256124e-05,
+      "loss": 0.301,
+      "step": 5451
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.728342749529191e-05,
+      "loss": 0.1669,
+      "step": 5452
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7281073446327685e-05,
+      "loss": 0.2227,
+      "step": 5453
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.727871939736346e-05,
+      "loss": 0.2232,
+      "step": 5454
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.727636534839925e-05,
+      "loss": 0.2004,
+      "step": 5455
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.727401129943503e-05,
+      "loss": 0.2182,
+      "step": 5456
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.727165725047081e-05,
+      "loss": 0.2332,
+      "step": 5457
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.72693032015066e-05,
+      "loss": 0.2289,
+      "step": 5458
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7266949152542375e-05,
+      "loss": 0.1961,
+      "step": 5459
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.726459510357815e-05,
+      "loss": 0.2085,
+      "step": 5460
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7262241054613936e-05,
+      "loss": 0.1556,
+      "step": 5461
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.725988700564972e-05,
+      "loss": 0.2826,
+      "step": 5462
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7257532956685504e-05,
+      "loss": 0.1873,
+      "step": 5463
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.725517890772128e-05,
+      "loss": 0.3259,
+      "step": 5464
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7252824858757065e-05,
+      "loss": 0.1767,
+      "step": 5465
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.725047080979285e-05,
+      "loss": 0.2283,
+      "step": 5466
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7248116760828626e-05,
+      "loss": 0.3878,
+      "step": 5467
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.724576271186441e-05,
+      "loss": 0.1692,
+      "step": 5468
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7243408662900194e-05,
+      "loss": 0.3692,
+      "step": 5469
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.724105461393597e-05,
+      "loss": 0.1916,
+      "step": 5470
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.723870056497175e-05,
+      "loss": 0.2781,
+      "step": 5471
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.723634651600754e-05,
+      "loss": 0.2543,
+      "step": 5472
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7233992467043316e-05,
+      "loss": 0.1342,
+      "step": 5473
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.723163841807909e-05,
+      "loss": 0.1164,
+      "step": 5474
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7229284369114884e-05,
+      "loss": 0.2202,
+      "step": 5475
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.722693032015066e-05,
+      "loss": 0.1599,
+      "step": 5476
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.722457627118644e-05,
+      "loss": 0.296,
+      "step": 5477
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.722222222222222e-05,
+      "loss": 0.2115,
+      "step": 5478
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7219868173258006e-05,
+      "loss": 0.29,
+      "step": 5479
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.721751412429379e-05,
+      "loss": 0.1624,
+      "step": 5480
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.721516007532957e-05,
+      "loss": 0.1921,
+      "step": 5481
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.721280602636535e-05,
+      "loss": 0.3249,
+      "step": 5482
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7210451977401135e-05,
+      "loss": 0.2256,
+      "step": 5483
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.720809792843691e-05,
+      "loss": 0.244,
+      "step": 5484
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7205743879472696e-05,
+      "loss": 0.2529,
+      "step": 5485
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.720338983050848e-05,
+      "loss": 0.1997,
+      "step": 5486
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.720103578154426e-05,
+      "loss": 0.2,
+      "step": 5487
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7198681732580034e-05,
+      "loss": 0.2004,
+      "step": 5488
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7196327683615825e-05,
+      "loss": 0.3074,
+      "step": 5489
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.71939736346516e-05,
+      "loss": 0.2144,
+      "step": 5490
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.719161958568738e-05,
+      "loss": 0.1624,
+      "step": 5491
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.718926553672317e-05,
+      "loss": 0.2707,
+      "step": 5492
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.718691148775895e-05,
+      "loss": 0.1952,
+      "step": 5493
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7184557438794724e-05,
+      "loss": 0.1619,
+      "step": 5494
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.718220338983051e-05,
+      "loss": 0.2764,
+      "step": 5495
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.717984934086629e-05,
+      "loss": 0.2102,
+      "step": 5496
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7177495291902076e-05,
+      "loss": 0.1381,
+      "step": 5497
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.717514124293785e-05,
+      "loss": 0.3198,
+      "step": 5498
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.717278719397364e-05,
+      "loss": 0.1623,
+      "step": 5499
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.717043314500942e-05,
+      "loss": 0.2217,
+      "step": 5500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.71680790960452e-05,
+      "loss": 0.1832,
+      "step": 5501
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.716572504708098e-05,
+      "loss": 0.3035,
+      "step": 5502
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7163370998116766e-05,
+      "loss": 0.2063,
+      "step": 5503
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.716101694915254e-05,
+      "loss": 0.3126,
+      "step": 5504
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.715866290018832e-05,
+      "loss": 0.1853,
+      "step": 5505
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.715630885122411e-05,
+      "loss": 0.2267,
+      "step": 5506
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.715395480225989e-05,
+      "loss": 0.1367,
+      "step": 5507
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7151600753295665e-05,
+      "loss": 0.2066,
+      "step": 5508
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7149246704331456e-05,
+      "loss": 0.3918,
+      "step": 5509
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.714689265536723e-05,
+      "loss": 0.2731,
+      "step": 5510
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.714453860640302e-05,
+      "loss": 0.2242,
+      "step": 5511
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7142184557438794e-05,
+      "loss": 0.3813,
+      "step": 5512
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.713983050847458e-05,
+      "loss": 0.2373,
+      "step": 5513
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.713747645951036e-05,
+      "loss": 0.2166,
+      "step": 5514
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.713512241054614e-05,
+      "loss": 0.1947,
+      "step": 5515
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.713276836158192e-05,
+      "loss": 0.1343,
+      "step": 5516
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.713041431261771e-05,
+      "loss": 0.2904,
+      "step": 5517
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7128060263653484e-05,
+      "loss": 0.2097,
+      "step": 5518
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.712570621468927e-05,
+      "loss": 0.2301,
+      "step": 5519
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.712335216572505e-05,
+      "loss": 0.1404,
+      "step": 5520
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.712099811676083e-05,
+      "loss": 0.0934,
+      "step": 5521
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.711864406779661e-05,
+      "loss": 0.2877,
+      "step": 5522
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.71162900188324e-05,
+      "loss": 0.1455,
+      "step": 5523
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7113935969868174e-05,
+      "loss": 0.412,
+      "step": 5524
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.711158192090395e-05,
+      "loss": 0.1326,
+      "step": 5525
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.710922787193974e-05,
+      "loss": 0.1427,
+      "step": 5526
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.710687382297552e-05,
+      "loss": 0.1807,
+      "step": 5527
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.71045197740113e-05,
+      "loss": 0.115,
+      "step": 5528
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.710216572504708e-05,
+      "loss": 0.18,
+      "step": 5529
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7099811676082864e-05,
+      "loss": 0.1355,
+      "step": 5530
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.709745762711865e-05,
+      "loss": 0.205,
+      "step": 5531
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7095103578154425e-05,
+      "loss": 0.3238,
+      "step": 5532
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.709274952919021e-05,
+      "loss": 0.2426,
+      "step": 5533
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.709039548022599e-05,
+      "loss": 0.2627,
+      "step": 5534
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.708804143126177e-05,
+      "loss": 0.2675,
+      "step": 5535
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7085687382297554e-05,
+      "loss": 0.1013,
+      "step": 5536
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.708333333333334e-05,
+      "loss": 0.1353,
+      "step": 5537
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7080979284369115e-05,
+      "loss": 0.2709,
+      "step": 5538
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.70786252354049e-05,
+      "loss": 0.2537,
+      "step": 5539
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.707627118644068e-05,
+      "loss": 0.2136,
+      "step": 5540
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.707391713747646e-05,
+      "loss": 0.2493,
+      "step": 5541
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7071563088512244e-05,
+      "loss": 0.1659,
+      "step": 5542
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.706920903954803e-05,
+      "loss": 0.1397,
+      "step": 5543
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7066854990583805e-05,
+      "loss": 0.342,
+      "step": 5544
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.706450094161959e-05,
+      "loss": 0.2141,
+      "step": 5545
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7062146892655366e-05,
+      "loss": 0.3241,
+      "step": 5546
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.705979284369115e-05,
+      "loss": 0.1828,
+      "step": 5547
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7057438794726934e-05,
+      "loss": 0.2988,
+      "step": 5548
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.705508474576271e-05,
+      "loss": 0.1652,
+      "step": 5549
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7052730696798495e-05,
+      "loss": 0.2238,
+      "step": 5550
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.705037664783428e-05,
+      "loss": 0.1092,
+      "step": 5551
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7048022598870056e-05,
+      "loss": 0.2749,
+      "step": 5552
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.704566854990584e-05,
+      "loss": 0.4008,
+      "step": 5553
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7043314500941624e-05,
+      "loss": 0.1568,
+      "step": 5554
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.70409604519774e-05,
+      "loss": 0.2449,
+      "step": 5555
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7038606403013185e-05,
+      "loss": 0.1577,
+      "step": 5556
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.703625235404897e-05,
+      "loss": 0.2423,
+      "step": 5557
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7033898305084746e-05,
+      "loss": 0.1203,
+      "step": 5558
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.703154425612053e-05,
+      "loss": 0.3356,
+      "step": 5559
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7029190207156314e-05,
+      "loss": 0.1236,
+      "step": 5560
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.702683615819209e-05,
+      "loss": 0.1365,
+      "step": 5561
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7024482109227875e-05,
+      "loss": 0.1159,
+      "step": 5562
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.702212806026365e-05,
+      "loss": 0.1944,
+      "step": 5563
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7019774011299436e-05,
+      "loss": 0.1821,
+      "step": 5564
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.701741996233522e-05,
+      "loss": 0.2168,
+      "step": 5565
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7015065913371e-05,
+      "loss": 0.2555,
+      "step": 5566
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.701271186440678e-05,
+      "loss": 0.0715,
+      "step": 5567
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.7010357815442565e-05,
+      "loss": 0.2259,
+      "step": 5568
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.700800376647834e-05,
+      "loss": 0.1602,
+      "step": 5569
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.7005649717514126e-05,
+      "loss": 0.3372,
+      "step": 5570
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.700329566854991e-05,
+      "loss": 0.3013,
+      "step": 5571
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.700094161958569e-05,
+      "loss": 0.17,
+      "step": 5572
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.699858757062147e-05,
+      "loss": 0.088,
+      "step": 5573
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6996233521657255e-05,
+      "loss": 0.324,
+      "step": 5574
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.699387947269303e-05,
+      "loss": 0.1545,
+      "step": 5575
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6991525423728816e-05,
+      "loss": 0.1543,
+      "step": 5576
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.69891713747646e-05,
+      "loss": 0.2375,
+      "step": 5577
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.698681732580038e-05,
+      "loss": 0.182,
+      "step": 5578
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.698446327683616e-05,
+      "loss": 0.069,
+      "step": 5579
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.698210922787194e-05,
+      "loss": 0.2071,
+      "step": 5580
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.697975517890772e-05,
+      "loss": 0.2044,
+      "step": 5581
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6977401129943506e-05,
+      "loss": 0.1845,
+      "step": 5582
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.697504708097928e-05,
+      "loss": 0.1987,
+      "step": 5583
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.697269303201507e-05,
+      "loss": 0.2007,
+      "step": 5584
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.697033898305085e-05,
+      "loss": 0.0742,
+      "step": 5585
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.696798493408663e-05,
+      "loss": 0.2911,
+      "step": 5586
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.696563088512241e-05,
+      "loss": 0.0534,
+      "step": 5587
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6963276836158196e-05,
+      "loss": 0.1871,
+      "step": 5588
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.696092278719397e-05,
+      "loss": 0.2434,
+      "step": 5589
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.695856873822976e-05,
+      "loss": 0.3192,
+      "step": 5590
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.695621468926554e-05,
+      "loss": 0.3195,
+      "step": 5591
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.695386064030132e-05,
+      "loss": 0.2292,
+      "step": 5592
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.69515065913371e-05,
+      "loss": 0.2126,
+      "step": 5593
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6949152542372886e-05,
+      "loss": 0.3008,
+      "step": 5594
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.694679849340866e-05,
+      "loss": 0.3117,
+      "step": 5595
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.694444444444445e-05,
+      "loss": 0.1484,
+      "step": 5596
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6942090395480224e-05,
+      "loss": 0.2004,
+      "step": 5597
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.693973634651601e-05,
+      "loss": 0.1105,
+      "step": 5598
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.693738229755179e-05,
+      "loss": 0.1947,
+      "step": 5599
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.693502824858757e-05,
+      "loss": 0.2825,
+      "step": 5600
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.693267419962335e-05,
+      "loss": 0.159,
+      "step": 5601
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.693032015065914e-05,
+      "loss": 0.1848,
+      "step": 5602
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6927966101694914e-05,
+      "loss": 0.201,
+      "step": 5603
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.69256120527307e-05,
+      "loss": 0.262,
+      "step": 5604
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.692325800376648e-05,
+      "loss": 0.1809,
+      "step": 5605
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.692090395480226e-05,
+      "loss": 0.3361,
+      "step": 5606
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.691854990583804e-05,
+      "loss": 0.2456,
+      "step": 5607
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.691619585687383e-05,
+      "loss": 0.2969,
+      "step": 5608
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6913841807909604e-05,
+      "loss": 0.1965,
+      "step": 5609
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.691148775894539e-05,
+      "loss": 0.1508,
+      "step": 5610
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.690913370998117e-05,
+      "loss": 0.1674,
+      "step": 5611
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.690677966101695e-05,
+      "loss": 0.2319,
+      "step": 5612
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.690442561205273e-05,
+      "loss": 0.176,
+      "step": 5613
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.690207156308852e-05,
+      "loss": 0.1225,
+      "step": 5614
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6899717514124294e-05,
+      "loss": 0.4038,
+      "step": 5615
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.689736346516008e-05,
+      "loss": 0.2916,
+      "step": 5616
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6895009416195855e-05,
+      "loss": 0.2814,
+      "step": 5617
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.689265536723164e-05,
+      "loss": 0.2591,
+      "step": 5618
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.689030131826742e-05,
+      "loss": 0.1045,
+      "step": 5619
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.68879472693032e-05,
+      "loss": 0.2245,
+      "step": 5620
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6885593220338984e-05,
+      "loss": 0.1277,
+      "step": 5621
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.688323917137477e-05,
+      "loss": 0.2268,
+      "step": 5622
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6880885122410545e-05,
+      "loss": 0.2868,
+      "step": 5623
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.687853107344633e-05,
+      "loss": 0.1356,
+      "step": 5624
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.687617702448211e-05,
+      "loss": 0.1244,
+      "step": 5625
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.687382297551789e-05,
+      "loss": 0.2271,
+      "step": 5626
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6871468926553674e-05,
+      "loss": 0.1149,
+      "step": 5627
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.686911487758946e-05,
+      "loss": 0.1668,
+      "step": 5628
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6866760828625235e-05,
+      "loss": 0.3851,
+      "step": 5629
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.686440677966102e-05,
+      "loss": 0.2581,
+      "step": 5630
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.68620527306968e-05,
+      "loss": 0.1583,
+      "step": 5631
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.685969868173258e-05,
+      "loss": 0.172,
+      "step": 5632
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6857344632768364e-05,
+      "loss": 0.124,
+      "step": 5633
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.685499058380414e-05,
+      "loss": 0.1447,
+      "step": 5634
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6852636534839925e-05,
+      "loss": 0.3057,
+      "step": 5635
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.685028248587571e-05,
+      "loss": 0.2956,
+      "step": 5636
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6847928436911486e-05,
+      "loss": 0.3108,
+      "step": 5637
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.684557438794727e-05,
+      "loss": 0.174,
+      "step": 5638
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6843220338983054e-05,
+      "loss": 0.2834,
+      "step": 5639
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.684086629001883e-05,
+      "loss": 0.1305,
+      "step": 5640
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6838512241054615e-05,
+      "loss": 0.2643,
+      "step": 5641
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.68361581920904e-05,
+      "loss": 0.1641,
+      "step": 5642
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6833804143126176e-05,
+      "loss": 0.3176,
+      "step": 5643
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.683145009416196e-05,
+      "loss": 0.1032,
+      "step": 5644
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6829096045197744e-05,
+      "loss": 0.2293,
+      "step": 5645
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.682674199623352e-05,
+      "loss": 0.1015,
+      "step": 5646
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6824387947269305e-05,
+      "loss": 0.3208,
+      "step": 5647
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.682203389830509e-05,
+      "loss": 0.1922,
+      "step": 5648
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6819679849340866e-05,
+      "loss": 0.191,
+      "step": 5649
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.681732580037665e-05,
+      "loss": 0.1576,
+      "step": 5650
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.681497175141243e-05,
+      "loss": 0.1752,
+      "step": 5651
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.681261770244821e-05,
+      "loss": 0.0663,
+      "step": 5652
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6810263653483995e-05,
+      "loss": 0.1787,
+      "step": 5653
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.680790960451977e-05,
+      "loss": 0.1406,
+      "step": 5654
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6805555555555556e-05,
+      "loss": 0.2839,
+      "step": 5655
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.680320150659134e-05,
+      "loss": 0.0817,
+      "step": 5656
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.680084745762712e-05,
+      "loss": 0.1859,
+      "step": 5657
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.67984934086629e-05,
+      "loss": 0.0812,
+      "step": 5658
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6796139359698685e-05,
+      "loss": 0.208,
+      "step": 5659
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.679378531073446e-05,
+      "loss": 0.133,
+      "step": 5660
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6791431261770246e-05,
+      "loss": 0.206,
+      "step": 5661
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.678907721280603e-05,
+      "loss": 0.2945,
+      "step": 5662
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.678672316384181e-05,
+      "loss": 0.1925,
+      "step": 5663
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.678436911487759e-05,
+      "loss": 0.2894,
+      "step": 5664
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6782015065913375e-05,
+      "loss": 0.249,
+      "step": 5665
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.677966101694915e-05,
+      "loss": 0.1823,
+      "step": 5666
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6777306967984936e-05,
+      "loss": 0.1985,
+      "step": 5667
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6774952919020714e-05,
+      "loss": 0.3369,
+      "step": 5668
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6772598870056504e-05,
+      "loss": 0.3327,
+      "step": 5669
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.677024482109228e-05,
+      "loss": 0.3265,
+      "step": 5670
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.676789077212806e-05,
+      "loss": 0.269,
+      "step": 5671
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.676553672316384e-05,
+      "loss": 0.4145,
+      "step": 5672
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6763182674199626e-05,
+      "loss": 0.119,
+      "step": 5673
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6760828625235404e-05,
+      "loss": 0.2618,
+      "step": 5674
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.675847457627119e-05,
+      "loss": 0.28,
+      "step": 5675
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.675612052730697e-05,
+      "loss": 0.3576,
+      "step": 5676
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.675376647834275e-05,
+      "loss": 0.181,
+      "step": 5677
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.675141242937853e-05,
+      "loss": 0.2519,
+      "step": 5678
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6749058380414316e-05,
+      "loss": 0.2308,
+      "step": 5679
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6746704331450094e-05,
+      "loss": 0.3583,
+      "step": 5680
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.674435028248588e-05,
+      "loss": 0.1839,
+      "step": 5681
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.674199623352166e-05,
+      "loss": 0.1241,
+      "step": 5682
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.673964218455744e-05,
+      "loss": 0.148,
+      "step": 5683
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.673728813559322e-05,
+      "loss": 0.175,
+      "step": 5684
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6734934086629e-05,
+      "loss": 0.2422,
+      "step": 5685
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.673258003766479e-05,
+      "loss": 0.3218,
+      "step": 5686
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.673022598870057e-05,
+      "loss": 0.1522,
+      "step": 5687
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6727871939736345e-05,
+      "loss": 0.2466,
+      "step": 5688
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6725517890772135e-05,
+      "loss": 0.1286,
+      "step": 5689
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.672316384180791e-05,
+      "loss": 0.2681,
+      "step": 5690
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.672080979284369e-05,
+      "loss": 0.2606,
+      "step": 5691
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6718455743879474e-05,
+      "loss": 0.164,
+      "step": 5692
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.671610169491526e-05,
+      "loss": 0.253,
+      "step": 5693
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6713747645951035e-05,
+      "loss": 0.2324,
+      "step": 5694
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.671139359698682e-05,
+      "loss": 0.1281,
+      "step": 5695
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.67090395480226e-05,
+      "loss": 0.2301,
+      "step": 5696
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.670668549905838e-05,
+      "loss": 0.1629,
+      "step": 5697
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6704331450094164e-05,
+      "loss": 0.2787,
+      "step": 5698
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.670197740112995e-05,
+      "loss": 0.0978,
+      "step": 5699
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.669962335216573e-05,
+      "loss": 0.1678,
+      "step": 5700
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.669726930320151e-05,
+      "loss": 0.2263,
+      "step": 5701
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6694915254237286e-05,
+      "loss": 0.3441,
+      "step": 5702
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6692561205273076e-05,
+      "loss": 0.1698,
+      "step": 5703
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6690207156308854e-05,
+      "loss": 0.1608,
+      "step": 5704
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.668785310734463e-05,
+      "loss": 0.2602,
+      "step": 5705
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.668549905838042e-05,
+      "loss": 0.2154,
+      "step": 5706
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.66831450094162e-05,
+      "loss": 0.1441,
+      "step": 5707
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6680790960451976e-05,
+      "loss": 0.2437,
+      "step": 5708
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.667843691148776e-05,
+      "loss": 0.1452,
+      "step": 5709
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6676082862523544e-05,
+      "loss": 0.218,
+      "step": 5710
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.667372881355932e-05,
+      "loss": 0.1459,
+      "step": 5711
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6671374764595105e-05,
+      "loss": 0.1499,
+      "step": 5712
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.666902071563089e-05,
+      "loss": 0.1931,
+      "step": 5713
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6666666666666666e-05,
+      "loss": 0.2095,
+      "step": 5714
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.666431261770245e-05,
+      "loss": 0.2722,
+      "step": 5715
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6661958568738234e-05,
+      "loss": 0.2199,
+      "step": 5716
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.665960451977402e-05,
+      "loss": 0.2173,
+      "step": 5717
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6657250470809795e-05,
+      "loss": 0.2151,
+      "step": 5718
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.665489642184557e-05,
+      "loss": 0.2148,
+      "step": 5719
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.665254237288136e-05,
+      "loss": 0.1577,
+      "step": 5720
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.665018832391714e-05,
+      "loss": 0.2944,
+      "step": 5721
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.664783427495292e-05,
+      "loss": 0.2654,
+      "step": 5722
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.664548022598871e-05,
+      "loss": 0.1962,
+      "step": 5723
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6643126177024485e-05,
+      "loss": 0.1183,
+      "step": 5724
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.664077212806026e-05,
+      "loss": 0.1972,
+      "step": 5725
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6638418079096046e-05,
+      "loss": 0.1341,
+      "step": 5726
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.663606403013183e-05,
+      "loss": 0.147,
+      "step": 5727
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.663370998116761e-05,
+      "loss": 0.1701,
+      "step": 5728
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.663135593220339e-05,
+      "loss": 0.2065,
+      "step": 5729
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6629001883239175e-05,
+      "loss": 0.2914,
+      "step": 5730
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.662664783427495e-05,
+      "loss": 0.2759,
+      "step": 5731
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6624293785310736e-05,
+      "loss": 0.2409,
+      "step": 5732
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.662193973634652e-05,
+      "loss": 0.1262,
+      "step": 5733
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6619585687382304e-05,
+      "loss": 0.258,
+      "step": 5734
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.661723163841808e-05,
+      "loss": 0.2115,
+      "step": 5735
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.661487758945386e-05,
+      "loss": 0.237,
+      "step": 5736
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.661252354048965e-05,
+      "loss": 0.1389,
+      "step": 5737
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6610169491525426e-05,
+      "loss": 0.146,
+      "step": 5738
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.66078154425612e-05,
+      "loss": 0.2439,
+      "step": 5739
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6605461393596994e-05,
+      "loss": 0.1982,
+      "step": 5740
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.660310734463277e-05,
+      "loss": 0.3949,
+      "step": 5741
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.660075329566855e-05,
+      "loss": 0.264,
+      "step": 5742
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.659839924670433e-05,
+      "loss": 0.1388,
+      "step": 5743
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6596045197740116e-05,
+      "loss": 0.3084,
+      "step": 5744
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.659369114877589e-05,
+      "loss": 0.1487,
+      "step": 5745
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.659133709981168e-05,
+      "loss": 0.2077,
+      "step": 5746
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.658898305084746e-05,
+      "loss": 0.3064,
+      "step": 5747
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6586629001883245e-05,
+      "loss": 0.2063,
+      "step": 5748
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.658427495291902e-05,
+      "loss": 0.2537,
+      "step": 5749
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6581920903954806e-05,
+      "loss": 0.2694,
+      "step": 5750
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.657956685499059e-05,
+      "loss": 0.2449,
+      "step": 5751
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.657721280602637e-05,
+      "loss": 0.1965,
+      "step": 5752
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6574858757062144e-05,
+      "loss": 0.1144,
+      "step": 5753
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6572504708097935e-05,
+      "loss": 0.1591,
+      "step": 5754
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.657015065913371e-05,
+      "loss": 0.2088,
+      "step": 5755
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.656779661016949e-05,
+      "loss": 0.3476,
+      "step": 5756
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.656544256120528e-05,
+      "loss": 0.2093,
+      "step": 5757
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.656308851224106e-05,
+      "loss": 0.1454,
+      "step": 5758
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6560734463276834e-05,
+      "loss": 0.2341,
+      "step": 5759
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.655838041431262e-05,
+      "loss": 0.1557,
+      "step": 5760
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.65560263653484e-05,
+      "loss": 0.3835,
+      "step": 5761
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.655367231638418e-05,
+      "loss": 0.231,
+      "step": 5762
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.655131826741996e-05,
+      "loss": 0.2463,
+      "step": 5763
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.654896421845575e-05,
+      "loss": 0.1397,
+      "step": 5764
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.654661016949153e-05,
+      "loss": 0.211,
+      "step": 5765
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.654425612052731e-05,
+      "loss": 0.2406,
+      "step": 5766
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.654190207156309e-05,
+      "loss": 0.3485,
+      "step": 5767
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6539548022598876e-05,
+      "loss": 0.0825,
+      "step": 5768
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.653719397363465e-05,
+      "loss": 0.2037,
+      "step": 5769
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.653483992467043e-05,
+      "loss": 0.1728,
+      "step": 5770
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.653248587570622e-05,
+      "loss": 0.3418,
+      "step": 5771
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6530131826742e-05,
+      "loss": 0.2103,
+      "step": 5772
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6527777777777775e-05,
+      "loss": 0.2298,
+      "step": 5773
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6525423728813566e-05,
+      "loss": 0.1594,
+      "step": 5774
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.652306967984934e-05,
+      "loss": 0.1267,
+      "step": 5775
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.652071563088512e-05,
+      "loss": 0.1826,
+      "step": 5776
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6518361581920904e-05,
+      "loss": 0.1691,
+      "step": 5777
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.651600753295669e-05,
+      "loss": 0.1445,
+      "step": 5778
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.651365348399247e-05,
+      "loss": 0.2009,
+      "step": 5779
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.651129943502825e-05,
+      "loss": 0.1458,
+      "step": 5780
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.650894538606403e-05,
+      "loss": 0.2186,
+      "step": 5781
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.650659133709982e-05,
+      "loss": 0.1448,
+      "step": 5782
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6504237288135594e-05,
+      "loss": 0.2841,
+      "step": 5783
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.650188323917138e-05,
+      "loss": 0.1265,
+      "step": 5784
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.649952919020716e-05,
+      "loss": 0.1882,
+      "step": 5785
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.649717514124294e-05,
+      "loss": 0.2191,
+      "step": 5786
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6494821092278716e-05,
+      "loss": 0.1736,
+      "step": 5787
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.649246704331451e-05,
+      "loss": 0.3546,
+      "step": 5788
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6490112994350284e-05,
+      "loss": 0.3684,
+      "step": 5789
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.648775894538606e-05,
+      "loss": 0.3432,
+      "step": 5790
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.648540489642185e-05,
+      "loss": 0.2014,
+      "step": 5791
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.648305084745763e-05,
+      "loss": 0.216,
+      "step": 5792
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6480696798493406e-05,
+      "loss": 0.1963,
+      "step": 5793
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.647834274952919e-05,
+      "loss": 0.2443,
+      "step": 5794
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6475988700564974e-05,
+      "loss": 0.178,
+      "step": 5795
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.647363465160076e-05,
+      "loss": 0.1966,
+      "step": 5796
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6471280602636535e-05,
+      "loss": 0.1787,
+      "step": 5797
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.646892655367232e-05,
+      "loss": 0.181,
+      "step": 5798
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.64665725047081e-05,
+      "loss": 0.1344,
+      "step": 5799
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.646421845574388e-05,
+      "loss": 0.3411,
+      "step": 5800
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6461864406779664e-05,
+      "loss": 0.184,
+      "step": 5801
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.645951035781545e-05,
+      "loss": 0.1417,
+      "step": 5802
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6457156308851225e-05,
+      "loss": 0.224,
+      "step": 5803
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6454802259887e-05,
+      "loss": 0.2768,
+      "step": 5804
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.645244821092279e-05,
+      "loss": 0.2806,
+      "step": 5805
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.645009416195857e-05,
+      "loss": 0.17,
+      "step": 5806
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.644774011299435e-05,
+      "loss": 0.2142,
+      "step": 5807
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.644538606403014e-05,
+      "loss": 0.2911,
+      "step": 5808
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6443032015065915e-05,
+      "loss": 0.0974,
+      "step": 5809
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.644067796610169e-05,
+      "loss": 0.1592,
+      "step": 5810
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6438323917137476e-05,
+      "loss": 0.1693,
+      "step": 5811
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.643596986817326e-05,
+      "loss": 0.2632,
+      "step": 5812
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6433615819209044e-05,
+      "loss": 0.2153,
+      "step": 5813
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.643126177024482e-05,
+      "loss": 0.1711,
+      "step": 5814
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6428907721280605e-05,
+      "loss": 0.1952,
+      "step": 5815
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.642655367231639e-05,
+      "loss": 0.1913,
+      "step": 5816
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6424199623352166e-05,
+      "loss": 0.1527,
+      "step": 5817
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.642184557438795e-05,
+      "loss": 0.191,
+      "step": 5818
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6419491525423734e-05,
+      "loss": 0.0949,
+      "step": 5819
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.641713747645951e-05,
+      "loss": 0.2971,
+      "step": 5820
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.641478342749529e-05,
+      "loss": 0.1164,
+      "step": 5821
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.641242937853108e-05,
+      "loss": 0.1491,
+      "step": 5822
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6410075329566856e-05,
+      "loss": 0.2409,
+      "step": 5823
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.640772128060263e-05,
+      "loss": 0.1074,
+      "step": 5824
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6405367231638424e-05,
+      "loss": 0.0862,
+      "step": 5825
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.64030131826742e-05,
+      "loss": 0.1655,
+      "step": 5826
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6400659133709985e-05,
+      "loss": 0.1026,
+      "step": 5827
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.639830508474576e-05,
+      "loss": 0.1243,
+      "step": 5828
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6395951035781546e-05,
+      "loss": 0.2298,
+      "step": 5829
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.639359698681733e-05,
+      "loss": 0.2164,
+      "step": 5830
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.639124293785311e-05,
+      "loss": 0.1701,
+      "step": 5831
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.638888888888889e-05,
+      "loss": 0.1804,
+      "step": 5832
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6386534839924675e-05,
+      "loss": 0.3629,
+      "step": 5833
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.638418079096045e-05,
+      "loss": 0.3811,
+      "step": 5834
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6381826741996236e-05,
+      "loss": 0.1478,
+      "step": 5835
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.637947269303202e-05,
+      "loss": 0.339,
+      "step": 5836
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.63771186440678e-05,
+      "loss": 0.1572,
+      "step": 5837
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.637476459510358e-05,
+      "loss": 0.1592,
+      "step": 5838
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6372410546139365e-05,
+      "loss": 0.0888,
+      "step": 5839
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.637005649717514e-05,
+      "loss": 0.2024,
+      "step": 5840
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.636770244821092e-05,
+      "loss": 0.2093,
+      "step": 5841
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.636534839924671e-05,
+      "loss": 0.1883,
+      "step": 5842
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.636299435028249e-05,
+      "loss": 0.2405,
+      "step": 5843
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.636064030131827e-05,
+      "loss": 0.1175,
+      "step": 5844
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.635828625235405e-05,
+      "loss": 0.1589,
+      "step": 5845
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.635593220338983e-05,
+      "loss": 0.1784,
+      "step": 5846
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6353578154425616e-05,
+      "loss": 0.4175,
+      "step": 5847
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.635122410546139e-05,
+      "loss": 0.1465,
+      "step": 5848
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.634887005649718e-05,
+      "loss": 0.1777,
+      "step": 5849
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.634651600753296e-05,
+      "loss": 0.2391,
+      "step": 5850
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.634416195856874e-05,
+      "loss": 0.3324,
+      "step": 5851
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.634180790960452e-05,
+      "loss": 0.2139,
+      "step": 5852
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6339453860640306e-05,
+      "loss": 0.1144,
+      "step": 5853
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.633709981167608e-05,
+      "loss": 0.2548,
+      "step": 5854
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.633474576271187e-05,
+      "loss": 0.2044,
+      "step": 5855
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.633239171374765e-05,
+      "loss": 0.1356,
+      "step": 5856
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.633003766478343e-05,
+      "loss": 0.169,
+      "step": 5857
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.632768361581921e-05,
+      "loss": 0.252,
+      "step": 5858
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6325329566854996e-05,
+      "loss": 0.2427,
+      "step": 5859
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.632297551789077e-05,
+      "loss": 0.2757,
+      "step": 5860
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.632062146892656e-05,
+      "loss": 0.1684,
+      "step": 5861
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6318267419962334e-05,
+      "loss": 0.247,
+      "step": 5862
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.631591337099812e-05,
+      "loss": 0.2045,
+      "step": 5863
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.63135593220339e-05,
+      "loss": 0.211,
+      "step": 5864
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.631120527306968e-05,
+      "loss": 0.2347,
+      "step": 5865
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.630885122410546e-05,
+      "loss": 0.327,
+      "step": 5866
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.630649717514125e-05,
+      "loss": 0.3634,
+      "step": 5867
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6304143126177024e-05,
+      "loss": 0.2136,
+      "step": 5868
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.630178907721281e-05,
+      "loss": 0.2252,
+      "step": 5869
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.629943502824859e-05,
+      "loss": 0.1947,
+      "step": 5870
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.629708097928437e-05,
+      "loss": 0.2616,
+      "step": 5871
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.629472693032015e-05,
+      "loss": 0.2209,
+      "step": 5872
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.629237288135594e-05,
+      "loss": 0.1507,
+      "step": 5873
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6290018832391714e-05,
+      "loss": 0.1652,
+      "step": 5874
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.62876647834275e-05,
+      "loss": 0.1217,
+      "step": 5875
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.628531073446328e-05,
+      "loss": 0.1965,
+      "step": 5876
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.628295668549906e-05,
+      "loss": 0.1783,
+      "step": 5877
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.628060263653484e-05,
+      "loss": 0.1569,
+      "step": 5878
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.627824858757062e-05,
+      "loss": 0.1125,
+      "step": 5879
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6275894538606404e-05,
+      "loss": 0.2401,
+      "step": 5880
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.627354048964219e-05,
+      "loss": 0.2726,
+      "step": 5881
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6271186440677965e-05,
+      "loss": 0.1795,
+      "step": 5882
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.626883239171375e-05,
+      "loss": 0.2381,
+      "step": 5883
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.626647834274953e-05,
+      "loss": 0.1781,
+      "step": 5884
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.626412429378531e-05,
+      "loss": 0.2824,
+      "step": 5885
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6261770244821094e-05,
+      "loss": 0.1831,
+      "step": 5886
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.625941619585688e-05,
+      "loss": 0.123,
+      "step": 5887
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6257062146892655e-05,
+      "loss": 0.1601,
+      "step": 5888
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.625470809792844e-05,
+      "loss": 0.0474,
+      "step": 5889
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.625235404896422e-05,
+      "loss": 0.3324,
+      "step": 5890
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.625e-05,
+      "loss": 0.2089,
+      "step": 5891
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6247645951035784e-05,
+      "loss": 0.2767,
+      "step": 5892
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.624529190207157e-05,
+      "loss": 0.2202,
+      "step": 5893
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6242937853107345e-05,
+      "loss": 0.3067,
+      "step": 5894
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.624058380414313e-05,
+      "loss": 0.1932,
+      "step": 5895
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6238229755178906e-05,
+      "loss": 0.1428,
+      "step": 5896
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.623587570621469e-05,
+      "loss": 0.2869,
+      "step": 5897
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6233521657250474e-05,
+      "loss": 0.1612,
+      "step": 5898
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.623116760828625e-05,
+      "loss": 0.2123,
+      "step": 5899
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6228813559322035e-05,
+      "loss": 0.2025,
+      "step": 5900
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.622645951035782e-05,
+      "loss": 0.1633,
+      "step": 5901
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6224105461393596e-05,
+      "loss": 0.1225,
+      "step": 5902
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.622175141242938e-05,
+      "loss": 0.2396,
+      "step": 5903
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6219397363465164e-05,
+      "loss": 0.2613,
+      "step": 5904
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.621704331450094e-05,
+      "loss": 0.0655,
+      "step": 5905
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6214689265536725e-05,
+      "loss": 0.1546,
+      "step": 5906
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.621233521657251e-05,
+      "loss": 0.2942,
+      "step": 5907
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6209981167608286e-05,
+      "loss": 0.1722,
+      "step": 5908
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.620762711864407e-05,
+      "loss": 0.1513,
+      "step": 5909
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6205273069679854e-05,
+      "loss": 0.1893,
+      "step": 5910
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.620291902071563e-05,
+      "loss": 0.2221,
+      "step": 5911
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6200564971751415e-05,
+      "loss": 0.3573,
+      "step": 5912
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.619821092278719e-05,
+      "loss": 0.1765,
+      "step": 5913
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6195856873822976e-05,
+      "loss": 0.1413,
+      "step": 5914
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.619350282485876e-05,
+      "loss": 0.2465,
+      "step": 5915
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.619114877589454e-05,
+      "loss": 0.1982,
+      "step": 5916
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.618879472693032e-05,
+      "loss": 0.1466,
+      "step": 5917
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6186440677966105e-05,
+      "loss": 0.1066,
+      "step": 5918
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.618408662900188e-05,
+      "loss": 0.3259,
+      "step": 5919
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6181732580037666e-05,
+      "loss": 0.1188,
+      "step": 5920
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.617937853107345e-05,
+      "loss": 0.3627,
+      "step": 5921
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.617702448210923e-05,
+      "loss": 0.1429,
+      "step": 5922
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.617467043314501e-05,
+      "loss": 0.2514,
+      "step": 5923
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6172316384180795e-05,
+      "loss": 0.2106,
+      "step": 5924
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.616996233521657e-05,
+      "loss": 0.2663,
+      "step": 5925
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6167608286252356e-05,
+      "loss": 0.2624,
+      "step": 5926
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.616525423728814e-05,
+      "loss": 0.2858,
+      "step": 5927
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.616290018832392e-05,
+      "loss": 0.195,
+      "step": 5928
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.61605461393597e-05,
+      "loss": 0.2197,
+      "step": 5929
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6158192090395485e-05,
+      "loss": 0.2373,
+      "step": 5930
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.615583804143126e-05,
+      "loss": 0.1998,
+      "step": 5931
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6153483992467046e-05,
+      "loss": 0.3029,
+      "step": 5932
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6151129943502823e-05,
+      "loss": 0.2062,
+      "step": 5933
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.614877589453861e-05,
+      "loss": 0.2584,
+      "step": 5934
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.614642184557439e-05,
+      "loss": 0.2059,
+      "step": 5935
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.614406779661017e-05,
+      "loss": 0.2936,
+      "step": 5936
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.614171374764595e-05,
+      "loss": 0.2117,
+      "step": 5937
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6139359698681736e-05,
+      "loss": 0.2459,
+      "step": 5938
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6137005649717513e-05,
+      "loss": 0.2618,
+      "step": 5939
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.61346516007533e-05,
+      "loss": 0.093,
+      "step": 5940
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.613229755178908e-05,
+      "loss": 0.1374,
+      "step": 5941
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.612994350282486e-05,
+      "loss": 0.1923,
+      "step": 5942
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.612758945386064e-05,
+      "loss": 0.3297,
+      "step": 5943
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6125235404896426e-05,
+      "loss": 0.2289,
+      "step": 5944
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.61228813559322e-05,
+      "loss": 0.1898,
+      "step": 5945
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.612052730696799e-05,
+      "loss": 0.1652,
+      "step": 5946
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.611817325800377e-05,
+      "loss": 0.1996,
+      "step": 5947
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.611581920903955e-05,
+      "loss": 0.1087,
+      "step": 5948
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.611346516007533e-05,
+      "loss": 0.1467,
+      "step": 5949
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.611111111111111e-05,
+      "loss": 0.1999,
+      "step": 5950
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.610875706214689e-05,
+      "loss": 0.1721,
+      "step": 5951
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.610640301318268e-05,
+      "loss": 0.248,
+      "step": 5952
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6104048964218454e-05,
+      "loss": 0.1838,
+      "step": 5953
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.610169491525424e-05,
+      "loss": 0.1375,
+      "step": 5954
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.609934086629002e-05,
+      "loss": 0.2113,
+      "step": 5955
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.60969868173258e-05,
+      "loss": 0.1545,
+      "step": 5956
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.609463276836158e-05,
+      "loss": 0.2528,
+      "step": 5957
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.609227871939737e-05,
+      "loss": 0.2751,
+      "step": 5958
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6089924670433144e-05,
+      "loss": 0.2657,
+      "step": 5959
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.608757062146893e-05,
+      "loss": 0.1783,
+      "step": 5960
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.608521657250471e-05,
+      "loss": 0.2891,
+      "step": 5961
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.608286252354049e-05,
+      "loss": 0.2264,
+      "step": 5962
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.608050847457627e-05,
+      "loss": 0.203,
+      "step": 5963
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.607815442561206e-05,
+      "loss": 0.2278,
+      "step": 5964
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6075800376647834e-05,
+      "loss": 0.3062,
+      "step": 5965
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.607344632768362e-05,
+      "loss": 0.0938,
+      "step": 5966
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6071092278719396e-05,
+      "loss": 0.1104,
+      "step": 5967
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.606873822975518e-05,
+      "loss": 0.309,
+      "step": 5968
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.606638418079096e-05,
+      "loss": 0.1579,
+      "step": 5969
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.606403013182674e-05,
+      "loss": 0.3622,
+      "step": 5970
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6061676082862524e-05,
+      "loss": 0.1475,
+      "step": 5971
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.605932203389831e-05,
+      "loss": 0.3009,
+      "step": 5972
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6056967984934086e-05,
+      "loss": 0.1597,
+      "step": 5973
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.605461393596987e-05,
+      "loss": 0.3338,
+      "step": 5974
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.605225988700565e-05,
+      "loss": 0.1284,
+      "step": 5975
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.604990583804143e-05,
+      "loss": 0.1772,
+      "step": 5976
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6047551789077214e-05,
+      "loss": 0.2527,
+      "step": 5977
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6045197740113e-05,
+      "loss": 0.1787,
+      "step": 5978
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6042843691148775e-05,
+      "loss": 0.2012,
+      "step": 5979
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.604048964218456e-05,
+      "loss": 0.2465,
+      "step": 5980
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.603813559322034e-05,
+      "loss": 0.259,
+      "step": 5981
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.603578154425612e-05,
+      "loss": 0.2342,
+      "step": 5982
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6033427495291904e-05,
+      "loss": 0.14,
+      "step": 5983
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.603107344632768e-05,
+      "loss": 0.2267,
+      "step": 5984
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.602871939736347e-05,
+      "loss": 0.1138,
+      "step": 5985
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.602636534839925e-05,
+      "loss": 0.2748,
+      "step": 5986
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6024011299435027e-05,
+      "loss": 0.189,
+      "step": 5987
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.602165725047081e-05,
+      "loss": 0.1557,
+      "step": 5988
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6019303201506594e-05,
+      "loss": 0.2601,
+      "step": 5989
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.601694915254237e-05,
+      "loss": 0.2639,
+      "step": 5990
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6014595103578155e-05,
+      "loss": 0.1634,
+      "step": 5991
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.601224105461394e-05,
+      "loss": 0.1731,
+      "step": 5992
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6009887005649717e-05,
+      "loss": 0.3462,
+      "step": 5993
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.60075329566855e-05,
+      "loss": 0.3962,
+      "step": 5994
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.6005178907721284e-05,
+      "loss": 0.1755,
+      "step": 5995
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.600282485875706e-05,
+      "loss": 0.1012,
+      "step": 5996
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.6000470809792845e-05,
+      "loss": 0.2002,
+      "step": 5997
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.599811676082863e-05,
+      "loss": 0.2022,
+      "step": 5998
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5995762711864407e-05,
+      "loss": 0.2273,
+      "step": 5999
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.599340866290019e-05,
+      "loss": 0.2854,
+      "step": 6000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.599105461393597e-05,
+      "loss": 0.3435,
+      "step": 6001
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.598870056497176e-05,
+      "loss": 0.1715,
+      "step": 6002
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5986346516007535e-05,
+      "loss": 0.115,
+      "step": 6003
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.598399246704331e-05,
+      "loss": 0.113,
+      "step": 6004
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.59816384180791e-05,
+      "loss": 0.1777,
+      "step": 6005
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.597928436911488e-05,
+      "loss": 0.1329,
+      "step": 6006
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.597693032015066e-05,
+      "loss": 0.2767,
+      "step": 6007
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.597457627118644e-05,
+      "loss": 0.2346,
+      "step": 6008
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5972222222222225e-05,
+      "loss": 0.2436,
+      "step": 6009
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5969868173258e-05,
+      "loss": 0.3378,
+      "step": 6010
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5967514124293787e-05,
+      "loss": 0.182,
+      "step": 6011
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.596516007532957e-05,
+      "loss": 0.1531,
+      "step": 6012
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.596280602636535e-05,
+      "loss": 0.2199,
+      "step": 6013
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.596045197740113e-05,
+      "loss": 0.3023,
+      "step": 6014
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5958097928436915e-05,
+      "loss": 0.2132,
+      "step": 6015
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.59557438794727e-05,
+      "loss": 0.1822,
+      "step": 6016
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5953389830508477e-05,
+      "loss": 0.2432,
+      "step": 6017
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5951035781544254e-05,
+      "loss": 0.2959,
+      "step": 6018
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5948681732580044e-05,
+      "loss": 0.2819,
+      "step": 6019
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.594632768361582e-05,
+      "loss": 0.2328,
+      "step": 6020
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.59439736346516e-05,
+      "loss": 0.1166,
+      "step": 6021
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.594161958568739e-05,
+      "loss": 0.1143,
+      "step": 6022
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5939265536723167e-05,
+      "loss": 0.2411,
+      "step": 6023
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5936911487758944e-05,
+      "loss": 0.1681,
+      "step": 6024
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.593455743879473e-05,
+      "loss": 0.1542,
+      "step": 6025
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.593220338983051e-05,
+      "loss": 0.1988,
+      "step": 6026
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.592984934086629e-05,
+      "loss": 0.1509,
+      "step": 6027
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.592749529190207e-05,
+      "loss": 0.1598,
+      "step": 6028
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5925141242937857e-05,
+      "loss": 0.0973,
+      "step": 6029
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5922787193973634e-05,
+      "loss": 0.2399,
+      "step": 6030
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.592043314500942e-05,
+      "loss": 0.1209,
+      "step": 6031
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.59180790960452e-05,
+      "loss": 0.0989,
+      "step": 6032
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5915725047080985e-05,
+      "loss": 0.1565,
+      "step": 6033
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.591337099811676e-05,
+      "loss": 0.2232,
+      "step": 6034
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.591101694915254e-05,
+      "loss": 0.1659,
+      "step": 6035
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.590866290018833e-05,
+      "loss": 0.3509,
+      "step": 6036
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.590630885122411e-05,
+      "loss": 0.0983,
+      "step": 6037
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5903954802259885e-05,
+      "loss": 0.1568,
+      "step": 6038
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5901600753295675e-05,
+      "loss": 0.2251,
+      "step": 6039
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.589924670433145e-05,
+      "loss": 0.1592,
+      "step": 6040
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.589689265536723e-05,
+      "loss": 0.1425,
+      "step": 6041
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5894538606403014e-05,
+      "loss": 0.1496,
+      "step": 6042
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.58921845574388e-05,
+      "loss": 0.2275,
+      "step": 6043
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5889830508474575e-05,
+      "loss": 0.1473,
+      "step": 6044
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.588747645951036e-05,
+      "loss": 0.0951,
+      "step": 6045
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.588512241054614e-05,
+      "loss": 0.3763,
+      "step": 6046
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.588276836158192e-05,
+      "loss": 0.2758,
+      "step": 6047
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5880414312617704e-05,
+      "loss": 0.1646,
+      "step": 6048
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.587806026365349e-05,
+      "loss": 0.3532,
+      "step": 6049
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.587570621468927e-05,
+      "loss": 0.1785,
+      "step": 6050
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.587335216572505e-05,
+      "loss": 0.2549,
+      "step": 6051
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5870998116760826e-05,
+      "loss": 0.1872,
+      "step": 6052
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5868644067796617e-05,
+      "loss": 0.2294,
+      "step": 6053
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5866290018832394e-05,
+      "loss": 0.2511,
+      "step": 6054
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.586393596986817e-05,
+      "loss": 0.2147,
+      "step": 6055
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.586158192090396e-05,
+      "loss": 0.1921,
+      "step": 6056
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.585922787193974e-05,
+      "loss": 0.1995,
+      "step": 6057
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5856873822975516e-05,
+      "loss": 0.1878,
+      "step": 6058
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.58545197740113e-05,
+      "loss": 0.3157,
+      "step": 6059
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5852165725047084e-05,
+      "loss": 0.1813,
+      "step": 6060
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.584981167608286e-05,
+      "loss": 0.1786,
+      "step": 6061
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5847457627118645e-05,
+      "loss": 0.265,
+      "step": 6062
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.584510357815443e-05,
+      "loss": 0.2799,
+      "step": 6063
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.584274952919021e-05,
+      "loss": 0.1545,
+      "step": 6064
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.584039548022599e-05,
+      "loss": 0.2487,
+      "step": 6065
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5838041431261774e-05,
+      "loss": 0.2292,
+      "step": 6066
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.583568738229756e-05,
+      "loss": 0.1189,
+      "step": 6067
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5833333333333335e-05,
+      "loss": 0.2181,
+      "step": 6068
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.583097928436911e-05,
+      "loss": 0.1541,
+      "step": 6069
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.58286252354049e-05,
+      "loss": 0.1233,
+      "step": 6070
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.582627118644068e-05,
+      "loss": 0.1988,
+      "step": 6071
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.582391713747646e-05,
+      "loss": 0.148,
+      "step": 6072
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.582156308851225e-05,
+      "loss": 0.2055,
+      "step": 6073
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5819209039548025e-05,
+      "loss": 0.2422,
+      "step": 6074
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.58168549905838e-05,
+      "loss": 0.1943,
+      "step": 6075
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5814500941619586e-05,
+      "loss": 0.2321,
+      "step": 6076
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.581214689265537e-05,
+      "loss": 0.0776,
+      "step": 6077
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.580979284369115e-05,
+      "loss": 0.1334,
+      "step": 6078
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.580743879472693e-05,
+      "loss": 0.1686,
+      "step": 6079
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5805084745762715e-05,
+      "loss": 0.1302,
+      "step": 6080
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.58027306967985e-05,
+      "loss": 0.2346,
+      "step": 6081
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5800376647834276e-05,
+      "loss": 0.0608,
+      "step": 6082
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.579802259887006e-05,
+      "loss": 0.1912,
+      "step": 6083
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5795668549905844e-05,
+      "loss": 0.1771,
+      "step": 6084
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.579331450094162e-05,
+      "loss": 0.1798,
+      "step": 6085
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.57909604519774e-05,
+      "loss": 0.1343,
+      "step": 6086
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.578860640301319e-05,
+      "loss": 0.2676,
+      "step": 6087
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5786252354048966e-05,
+      "loss": 0.123,
+      "step": 6088
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.578389830508474e-05,
+      "loss": 0.1334,
+      "step": 6089
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5781544256120534e-05,
+      "loss": 0.0762,
+      "step": 6090
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.577919020715631e-05,
+      "loss": 0.1913,
+      "step": 6091
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.577683615819209e-05,
+      "loss": 0.1187,
+      "step": 6092
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.577448210922787e-05,
+      "loss": 0.3334,
+      "step": 6093
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5772128060263656e-05,
+      "loss": 0.0915,
+      "step": 6094
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.576977401129944e-05,
+      "loss": 0.2761,
+      "step": 6095
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.576741996233522e-05,
+      "loss": 0.2586,
+      "step": 6096
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5765065913371e-05,
+      "loss": 0.1106,
+      "step": 6097
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5762711864406785e-05,
+      "loss": 0.1306,
+      "step": 6098
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.576035781544256e-05,
+      "loss": 0.1109,
+      "step": 6099
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5758003766478346e-05,
+      "loss": 0.1318,
+      "step": 6100
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.575564971751413e-05,
+      "loss": 0.1759,
+      "step": 6101
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.575329566854991e-05,
+      "loss": 0.3266,
+      "step": 6102
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5750941619585684e-05,
+      "loss": 0.1308,
+      "step": 6103
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5748587570621475e-05,
+      "loss": 0.0987,
+      "step": 6104
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.574623352165725e-05,
+      "loss": 0.2278,
+      "step": 6105
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.574387947269303e-05,
+      "loss": 0.25,
+      "step": 6106
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.574152542372882e-05,
+      "loss": 0.0901,
+      "step": 6107
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.57391713747646e-05,
+      "loss": 0.2601,
+      "step": 6108
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5736817325800374e-05,
+      "loss": 0.2485,
+      "step": 6109
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.573446327683616e-05,
+      "loss": 0.2648,
+      "step": 6110
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.573210922787194e-05,
+      "loss": 0.2145,
+      "step": 6111
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5729755178907726e-05,
+      "loss": 0.1654,
+      "step": 6112
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.57274011299435e-05,
+      "loss": 0.276,
+      "step": 6113
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.572504708097929e-05,
+      "loss": 0.2199,
+      "step": 6114
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.572269303201507e-05,
+      "loss": 0.2409,
+      "step": 6115
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.572033898305085e-05,
+      "loss": 0.1955,
+      "step": 6116
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.571798493408663e-05,
+      "loss": 0.1896,
+      "step": 6117
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5715630885122416e-05,
+      "loss": 0.1738,
+      "step": 6118
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.571327683615819e-05,
+      "loss": 0.2095,
+      "step": 6119
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.571092278719397e-05,
+      "loss": 0.1932,
+      "step": 6120
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.570856873822976e-05,
+      "loss": 0.295,
+      "step": 6121
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.570621468926554e-05,
+      "loss": 0.1918,
+      "step": 6122
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5703860640301315e-05,
+      "loss": 0.1958,
+      "step": 6123
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5701506591337106e-05,
+      "loss": 0.1666,
+      "step": 6124
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.569915254237288e-05,
+      "loss": 0.1911,
+      "step": 6125
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.569679849340866e-05,
+      "loss": 0.3475,
+      "step": 6126
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5694444444444444e-05,
+      "loss": 0.1727,
+      "step": 6127
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.569209039548023e-05,
+      "loss": 0.2522,
+      "step": 6128
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.568973634651601e-05,
+      "loss": 0.1759,
+      "step": 6129
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.568738229755179e-05,
+      "loss": 0.2567,
+      "step": 6130
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.568502824858757e-05,
+      "loss": 0.1655,
+      "step": 6131
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.568267419962336e-05,
+      "loss": 0.1851,
+      "step": 6132
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5680320150659134e-05,
+      "loss": 0.2914,
+      "step": 6133
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.567796610169492e-05,
+      "loss": 0.1847,
+      "step": 6134
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.56756120527307e-05,
+      "loss": 0.1598,
+      "step": 6135
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.567325800376648e-05,
+      "loss": 0.1303,
+      "step": 6136
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5670903954802256e-05,
+      "loss": 0.2149,
+      "step": 6137
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.566854990583805e-05,
+      "loss": 0.1476,
+      "step": 6138
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5666195856873824e-05,
+      "loss": 0.1533,
+      "step": 6139
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.56638418079096e-05,
+      "loss": 0.2819,
+      "step": 6140
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.566148775894539e-05,
+      "loss": 0.278,
+      "step": 6141
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.565913370998117e-05,
+      "loss": 0.4113,
+      "step": 6142
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.565677966101695e-05,
+      "loss": 0.1909,
+      "step": 6143
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.565442561205273e-05,
+      "loss": 0.2187,
+      "step": 6144
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5652071563088514e-05,
+      "loss": 0.2741,
+      "step": 6145
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.56497175141243e-05,
+      "loss": 0.1773,
+      "step": 6146
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5647363465160075e-05,
+      "loss": 0.3246,
+      "step": 6147
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.564500941619586e-05,
+      "loss": 0.1564,
+      "step": 6148
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.564265536723164e-05,
+      "loss": 0.2478,
+      "step": 6149
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.564030131826742e-05,
+      "loss": 0.1947,
+      "step": 6150
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5637947269303204e-05,
+      "loss": 0.209,
+      "step": 6151
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.563559322033899e-05,
+      "loss": 0.2237,
+      "step": 6152
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5633239171374765e-05,
+      "loss": 0.3578,
+      "step": 6153
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.563088512241055e-05,
+      "loss": 0.1116,
+      "step": 6154
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.562853107344633e-05,
+      "loss": 0.242,
+      "step": 6155
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.562617702448211e-05,
+      "loss": 0.2555,
+      "step": 6156
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.562382297551789e-05,
+      "loss": 0.2135,
+      "step": 6157
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.562146892655368e-05,
+      "loss": 0.2161,
+      "step": 6158
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5619114877589455e-05,
+      "loss": 0.1771,
+      "step": 6159
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.561676082862524e-05,
+      "loss": 0.202,
+      "step": 6160
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5614406779661016e-05,
+      "loss": 0.328,
+      "step": 6161
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.56120527306968e-05,
+      "loss": 0.2748,
+      "step": 6162
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5609698681732584e-05,
+      "loss": 0.1423,
+      "step": 6163
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.560734463276836e-05,
+      "loss": 0.144,
+      "step": 6164
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5604990583804145e-05,
+      "loss": 0.1996,
+      "step": 6165
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.560263653483993e-05,
+      "loss": 0.1847,
+      "step": 6166
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5600282485875706e-05,
+      "loss": 0.231,
+      "step": 6167
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.559792843691149e-05,
+      "loss": 0.2722,
+      "step": 6168
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5595574387947274e-05,
+      "loss": 0.2267,
+      "step": 6169
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.559322033898305e-05,
+      "loss": 0.1686,
+      "step": 6170
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5590866290018835e-05,
+      "loss": 0.126,
+      "step": 6171
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.558851224105462e-05,
+      "loss": 0.1867,
+      "step": 6172
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5586158192090396e-05,
+      "loss": 0.1176,
+      "step": 6173
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.558380414312618e-05,
+      "loss": 0.1325,
+      "step": 6174
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5581450094161964e-05,
+      "loss": 0.281,
+      "step": 6175
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.557909604519774e-05,
+      "loss": 0.2004,
+      "step": 6176
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5576741996233525e-05,
+      "loss": 0.1033,
+      "step": 6177
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.55743879472693e-05,
+      "loss": 0.1741,
+      "step": 6178
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5572033898305086e-05,
+      "loss": 0.182,
+      "step": 6179
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.556967984934087e-05,
+      "loss": 0.217,
+      "step": 6180
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.556732580037665e-05,
+      "loss": 0.3071,
+      "step": 6181
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.556497175141243e-05,
+      "loss": 0.3245,
+      "step": 6182
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5562617702448215e-05,
+      "loss": 0.242,
+      "step": 6183
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.556026365348399e-05,
+      "loss": 0.2108,
+      "step": 6184
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5557909604519776e-05,
+      "loss": 0.1671,
+      "step": 6185
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.555555555555556e-05,
+      "loss": 0.2043,
+      "step": 6186
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.555320150659134e-05,
+      "loss": 0.3454,
+      "step": 6187
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.555084745762712e-05,
+      "loss": 0.166,
+      "step": 6188
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5548493408662905e-05,
+      "loss": 0.2008,
+      "step": 6189
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.554613935969868e-05,
+      "loss": 0.2491,
+      "step": 6190
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5543785310734466e-05,
+      "loss": 0.4284,
+      "step": 6191
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.554143126177025e-05,
+      "loss": 0.288,
+      "step": 6192
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.553907721280603e-05,
+      "loss": 0.1228,
+      "step": 6193
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.553672316384181e-05,
+      "loss": 0.2645,
+      "step": 6194
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.553436911487759e-05,
+      "loss": 0.197,
+      "step": 6195
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.553201506591337e-05,
+      "loss": 0.1788,
+      "step": 6196
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5529661016949156e-05,
+      "loss": 0.09,
+      "step": 6197
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.552730696798493e-05,
+      "loss": 0.1561,
+      "step": 6198
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.552495291902072e-05,
+      "loss": 0.1854,
+      "step": 6199
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.55225988700565e-05,
+      "loss": 0.1784,
+      "step": 6200
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.552024482109228e-05,
+      "loss": 0.0999,
+      "step": 6201
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.551789077212806e-05,
+      "loss": 0.2718,
+      "step": 6202
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5515536723163846e-05,
+      "loss": 0.209,
+      "step": 6203
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.551318267419962e-05,
+      "loss": 0.2526,
+      "step": 6204
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.551082862523541e-05,
+      "loss": 0.2476,
+      "step": 6205
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.550847457627119e-05,
+      "loss": 0.1859,
+      "step": 6206
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.550612052730697e-05,
+      "loss": 0.1447,
+      "step": 6207
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.550376647834275e-05,
+      "loss": 0.1613,
+      "step": 6208
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5501412429378536e-05,
+      "loss": 0.2626,
+      "step": 6209
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.549905838041431e-05,
+      "loss": 0.4111,
+      "step": 6210
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.54967043314501e-05,
+      "loss": 0.2021,
+      "step": 6211
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5494350282485874e-05,
+      "loss": 0.2191,
+      "step": 6212
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.549199623352166e-05,
+      "loss": 0.3457,
+      "step": 6213
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.548964218455744e-05,
+      "loss": 0.2172,
+      "step": 6214
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.548728813559322e-05,
+      "loss": 0.1406,
+      "step": 6215
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5484934086629e-05,
+      "loss": 0.3155,
+      "step": 6216
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.548258003766479e-05,
+      "loss": 0.2613,
+      "step": 6217
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5480225988700564e-05,
+      "loss": 0.1404,
+      "step": 6218
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.547787193973635e-05,
+      "loss": 0.2384,
+      "step": 6219
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.547551789077213e-05,
+      "loss": 0.2008,
+      "step": 6220
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.547316384180791e-05,
+      "loss": 0.1324,
+      "step": 6221
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.547080979284369e-05,
+      "loss": 0.1562,
+      "step": 6222
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.546845574387948e-05,
+      "loss": 0.2235,
+      "step": 6223
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5466101694915254e-05,
+      "loss": 0.3136,
+      "step": 6224
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.546374764595104e-05,
+      "loss": 0.1603,
+      "step": 6225
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.546139359698682e-05,
+      "loss": 0.2303,
+      "step": 6226
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.54590395480226e-05,
+      "loss": 0.3041,
+      "step": 6227
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.545668549905838e-05,
+      "loss": 0.2129,
+      "step": 6228
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.545433145009416e-05,
+      "loss": 0.2546,
+      "step": 6229
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5451977401129944e-05,
+      "loss": 0.1535,
+      "step": 6230
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.544962335216573e-05,
+      "loss": 0.1277,
+      "step": 6231
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5447269303201505e-05,
+      "loss": 0.2352,
+      "step": 6232
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.544491525423729e-05,
+      "loss": 0.1299,
+      "step": 6233
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.544256120527307e-05,
+      "loss": 0.2911,
+      "step": 6234
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.544020715630885e-05,
+      "loss": 0.1532,
+      "step": 6235
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5437853107344634e-05,
+      "loss": 0.2156,
+      "step": 6236
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.543549905838042e-05,
+      "loss": 0.0848,
+      "step": 6237
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5433145009416195e-05,
+      "loss": 0.1439,
+      "step": 6238
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.543079096045198e-05,
+      "loss": 0.2742,
+      "step": 6239
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.542843691148776e-05,
+      "loss": 0.2997,
+      "step": 6240
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.542608286252354e-05,
+      "loss": 0.1881,
+      "step": 6241
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5423728813559324e-05,
+      "loss": 0.1233,
+      "step": 6242
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.542137476459511e-05,
+      "loss": 0.1854,
+      "step": 6243
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5419020715630885e-05,
+      "loss": 0.1406,
+      "step": 6244
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.541666666666667e-05,
+      "loss": 0.2061,
+      "step": 6245
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.541431261770245e-05,
+      "loss": 0.1989,
+      "step": 6246
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.541195856873823e-05,
+      "loss": 0.1893,
+      "step": 6247
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5409604519774014e-05,
+      "loss": 0.2248,
+      "step": 6248
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.540725047080979e-05,
+      "loss": 0.2196,
+      "step": 6249
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5404896421845575e-05,
+      "loss": 0.0494,
+      "step": 6250
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.540254237288136e-05,
+      "loss": 0.2525,
+      "step": 6251
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5400188323917136e-05,
+      "loss": 0.2189,
+      "step": 6252
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.539783427495292e-05,
+      "loss": 0.1909,
+      "step": 6253
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5395480225988704e-05,
+      "loss": 0.1813,
+      "step": 6254
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.539312617702448e-05,
+      "loss": 0.1269,
+      "step": 6255
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5390772128060265e-05,
+      "loss": 0.1092,
+      "step": 6256
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.538841807909605e-05,
+      "loss": 0.2385,
+      "step": 6257
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5386064030131826e-05,
+      "loss": 0.1239,
+      "step": 6258
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.538370998116761e-05,
+      "loss": 0.1854,
+      "step": 6259
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5381355932203394e-05,
+      "loss": 0.1853,
+      "step": 6260
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.537900188323917e-05,
+      "loss": 0.2493,
+      "step": 6261
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5376647834274955e-05,
+      "loss": 0.1922,
+      "step": 6262
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.537429378531074e-05,
+      "loss": 0.2089,
+      "step": 6263
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5371939736346516e-05,
+      "loss": 0.2174,
+      "step": 6264
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.53695856873823e-05,
+      "loss": 0.3428,
+      "step": 6265
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.536723163841808e-05,
+      "loss": 0.156,
+      "step": 6266
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.536487758945386e-05,
+      "loss": 0.2387,
+      "step": 6267
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5362523540489645e-05,
+      "loss": 0.2578,
+      "step": 6268
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.536016949152542e-05,
+      "loss": 0.237,
+      "step": 6269
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5357815442561206e-05,
+      "loss": 0.2045,
+      "step": 6270
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.535546139359699e-05,
+      "loss": 0.2026,
+      "step": 6271
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.535310734463277e-05,
+      "loss": 0.2088,
+      "step": 6272
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.535075329566855e-05,
+      "loss": 0.0852,
+      "step": 6273
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5348399246704335e-05,
+      "loss": 0.1651,
+      "step": 6274
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.534604519774011e-05,
+      "loss": 0.2911,
+      "step": 6275
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5343691148775896e-05,
+      "loss": 0.1071,
+      "step": 6276
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.534133709981168e-05,
+      "loss": 0.1668,
+      "step": 6277
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.533898305084746e-05,
+      "loss": 0.1817,
+      "step": 6278
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.533662900188324e-05,
+      "loss": 0.212,
+      "step": 6279
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5334274952919025e-05,
+      "loss": 0.3027,
+      "step": 6280
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.53319209039548e-05,
+      "loss": 0.1665,
+      "step": 6281
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5329566854990586e-05,
+      "loss": 0.2548,
+      "step": 6282
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5327212806026363e-05,
+      "loss": 0.2632,
+      "step": 6283
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.532485875706215e-05,
+      "loss": 0.1553,
+      "step": 6284
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.532250470809793e-05,
+      "loss": 0.0959,
+      "step": 6285
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.532015065913371e-05,
+      "loss": 0.0944,
+      "step": 6286
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.531779661016949e-05,
+      "loss": 0.4003,
+      "step": 6287
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5315442561205276e-05,
+      "loss": 0.194,
+      "step": 6288
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5313088512241053e-05,
+      "loss": 0.1977,
+      "step": 6289
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.531073446327684e-05,
+      "loss": 0.1931,
+      "step": 6290
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.530838041431262e-05,
+      "loss": 0.2174,
+      "step": 6291
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.53060263653484e-05,
+      "loss": 0.0902,
+      "step": 6292
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.530367231638418e-05,
+      "loss": 0.1934,
+      "step": 6293
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5301318267419966e-05,
+      "loss": 0.2149,
+      "step": 6294
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5298964218455743e-05,
+      "loss": 0.1153,
+      "step": 6295
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.529661016949153e-05,
+      "loss": 0.0963,
+      "step": 6296
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.529425612052731e-05,
+      "loss": 0.2268,
+      "step": 6297
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.529190207156309e-05,
+      "loss": 0.1932,
+      "step": 6298
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.528954802259887e-05,
+      "loss": 0.2891,
+      "step": 6299
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.528719397363465e-05,
+      "loss": 0.1738,
+      "step": 6300
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.528483992467044e-05,
+      "loss": 0.2634,
+      "step": 6301
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.528248587570622e-05,
+      "loss": 0.1538,
+      "step": 6302
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5280131826741995e-05,
+      "loss": 0.2429,
+      "step": 6303
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.527777777777778e-05,
+      "loss": 0.2804,
+      "step": 6304
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.527542372881356e-05,
+      "loss": 0.2658,
+      "step": 6305
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.527306967984934e-05,
+      "loss": 0.1278,
+      "step": 6306
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5270715630885123e-05,
+      "loss": 0.2446,
+      "step": 6307
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.526836158192091e-05,
+      "loss": 0.1647,
+      "step": 6308
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5266007532956685e-05,
+      "loss": 0.2754,
+      "step": 6309
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.526365348399247e-05,
+      "loss": 0.1479,
+      "step": 6310
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.526129943502825e-05,
+      "loss": 0.2132,
+      "step": 6311
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.525894538606403e-05,
+      "loss": 0.1489,
+      "step": 6312
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5256591337099813e-05,
+      "loss": 0.2531,
+      "step": 6313
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.52542372881356e-05,
+      "loss": 0.1585,
+      "step": 6314
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5251883239171375e-05,
+      "loss": 0.1536,
+      "step": 6315
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.524952919020716e-05,
+      "loss": 0.2002,
+      "step": 6316
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5247175141242936e-05,
+      "loss": 0.0736,
+      "step": 6317
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5244821092278726e-05,
+      "loss": 0.124,
+      "step": 6318
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5242467043314503e-05,
+      "loss": 0.1608,
+      "step": 6319
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.524011299435028e-05,
+      "loss": 0.1251,
+      "step": 6320
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.523775894538607e-05,
+      "loss": 0.2243,
+      "step": 6321
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.523540489642185e-05,
+      "loss": 0.2,
+      "step": 6322
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5233050847457626e-05,
+      "loss": 0.1581,
+      "step": 6323
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.523069679849341e-05,
+      "loss": 0.1589,
+      "step": 6324
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5228342749529193e-05,
+      "loss": 0.211,
+      "step": 6325
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.522598870056497e-05,
+      "loss": 0.2409,
+      "step": 6326
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5223634651600755e-05,
+      "loss": 0.1615,
+      "step": 6327
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.522128060263654e-05,
+      "loss": 0.221,
+      "step": 6328
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5218926553672316e-05,
+      "loss": 0.2826,
+      "step": 6329
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.52165725047081e-05,
+      "loss": 0.1913,
+      "step": 6330
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5214218455743883e-05,
+      "loss": 0.2829,
+      "step": 6331
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.521186440677967e-05,
+      "loss": 0.2351,
+      "step": 6332
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5209510357815444e-05,
+      "loss": 0.1804,
+      "step": 6333
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.520715630885122e-05,
+      "loss": 0.2065,
+      "step": 6334
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.520480225988701e-05,
+      "loss": 0.1123,
+      "step": 6335
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.520244821092279e-05,
+      "loss": 0.1579,
+      "step": 6336
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.520009416195857e-05,
+      "loss": 0.2388,
+      "step": 6337
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.519774011299436e-05,
+      "loss": 0.1831,
+      "step": 6338
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5195386064030134e-05,
+      "loss": 0.1567,
+      "step": 6339
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.519303201506591e-05,
+      "loss": 0.1571,
+      "step": 6340
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5190677966101696e-05,
+      "loss": 0.2481,
+      "step": 6341
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.518832391713748e-05,
+      "loss": 0.2124,
+      "step": 6342
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.518596986817326e-05,
+      "loss": 0.2168,
+      "step": 6343
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.518361581920904e-05,
+      "loss": 0.2074,
+      "step": 6344
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5181261770244824e-05,
+      "loss": 0.1569,
+      "step": 6345
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.51789077212806e-05,
+      "loss": 0.256,
+      "step": 6346
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5176553672316386e-05,
+      "loss": 0.2191,
+      "step": 6347
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.517419962335217e-05,
+      "loss": 0.3162,
+      "step": 6348
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5171845574387953e-05,
+      "loss": 0.3188,
+      "step": 6349
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.516949152542373e-05,
+      "loss": 0.1927,
+      "step": 6350
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.516713747645951e-05,
+      "loss": 0.1989,
+      "step": 6351
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.51647834274953e-05,
+      "loss": 0.2086,
+      "step": 6352
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5162429378531076e-05,
+      "loss": 0.2818,
+      "step": 6353
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.516007532956685e-05,
+      "loss": 0.2453,
+      "step": 6354
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5157721280602643e-05,
+      "loss": 0.3695,
+      "step": 6355
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.515536723163842e-05,
+      "loss": 0.1981,
+      "step": 6356
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.51530131826742e-05,
+      "loss": 0.2364,
+      "step": 6357
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.515065913370998e-05,
+      "loss": 0.1365,
+      "step": 6358
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5148305084745766e-05,
+      "loss": 0.2112,
+      "step": 6359
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.514595103578154e-05,
+      "loss": 0.1955,
+      "step": 6360
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.514359698681733e-05,
+      "loss": 0.2259,
+      "step": 6361
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.514124293785311e-05,
+      "loss": 0.322,
+      "step": 6362
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.513888888888889e-05,
+      "loss": 0.1729,
+      "step": 6363
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.513653483992467e-05,
+      "loss": 0.2221,
+      "step": 6364
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5134180790960456e-05,
+      "loss": 0.2033,
+      "step": 6365
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.513182674199624e-05,
+      "loss": 0.2584,
+      "step": 6366
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.512947269303202e-05,
+      "loss": 0.1893,
+      "step": 6367
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5127118644067794e-05,
+      "loss": 0.1103,
+      "step": 6368
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5124764595103584e-05,
+      "loss": 0.1537,
+      "step": 6369
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.512241054613936e-05,
+      "loss": 0.1746,
+      "step": 6370
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.512005649717514e-05,
+      "loss": 0.2683,
+      "step": 6371
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.511770244821093e-05,
+      "loss": 0.2413,
+      "step": 6372
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5115348399246707e-05,
+      "loss": 0.2484,
+      "step": 6373
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5112994350282484e-05,
+      "loss": 0.2462,
+      "step": 6374
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.511064030131827e-05,
+      "loss": 0.2974,
+      "step": 6375
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.510828625235405e-05,
+      "loss": 0.2254,
+      "step": 6376
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.510593220338983e-05,
+      "loss": 0.1369,
+      "step": 6377
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.510357815442561e-05,
+      "loss": 0.3052,
+      "step": 6378
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5101224105461397e-05,
+      "loss": 0.1749,
+      "step": 6379
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.509887005649718e-05,
+      "loss": 0.2639,
+      "step": 6380
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.509651600753296e-05,
+      "loss": 0.2379,
+      "step": 6381
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.509416195856874e-05,
+      "loss": 0.2745,
+      "step": 6382
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5091807909604526e-05,
+      "loss": 0.0961,
+      "step": 6383
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.50894538606403e-05,
+      "loss": 0.217,
+      "step": 6384
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.508709981167608e-05,
+      "loss": 0.2325,
+      "step": 6385
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.508474576271187e-05,
+      "loss": 0.399,
+      "step": 6386
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.508239171374765e-05,
+      "loss": 0.2367,
+      "step": 6387
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.8496755058291486,
+      "eval_loss": 0.40403881669044495,
+      "eval_runtime": 22.307,
+      "eval_samples_per_second": 1526.563,
+      "eval_steps_per_second": 23.894,
+      "step": 6387
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5080037664783425e-05,
+      "loss": 0.1286,
+      "step": 6388
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5077683615819216e-05,
+      "loss": 0.1455,
+      "step": 6389
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.507532956685499e-05,
+      "loss": 0.1025,
+      "step": 6390
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.507297551789077e-05,
+      "loss": 0.1376,
+      "step": 6391
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5070621468926554e-05,
+      "loss": 0.1246,
+      "step": 6392
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.506826741996234e-05,
+      "loss": 0.1203,
+      "step": 6393
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5065913370998115e-05,
+      "loss": 0.1368,
+      "step": 6394
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.50635593220339e-05,
+      "loss": 0.0535,
+      "step": 6395
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.506120527306968e-05,
+      "loss": 0.137,
+      "step": 6396
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5058851224105467e-05,
+      "loss": 0.1489,
+      "step": 6397
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.5056497175141244e-05,
+      "loss": 0.2267,
+      "step": 6398
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.505414312617703e-05,
+      "loss": 0.2502,
+      "step": 6399
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.505178907721281e-05,
+      "loss": 0.0617,
+      "step": 6400
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.504943502824859e-05,
+      "loss": 0.2323,
+      "step": 6401
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.5047080979284366e-05,
+      "loss": 0.1335,
+      "step": 6402
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.5044726930320157e-05,
+      "loss": 0.1565,
+      "step": 6403
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.5042372881355934e-05,
+      "loss": 0.1342,
+      "step": 6404
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.504001883239171e-05,
+      "loss": 0.2021,
+      "step": 6405
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.50376647834275e-05,
+      "loss": 0.0604,
+      "step": 6406
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.503531073446328e-05,
+      "loss": 0.1166,
+      "step": 6407
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.5032956685499056e-05,
+      "loss": 0.1789,
+      "step": 6408
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.503060263653484e-05,
+      "loss": 0.2013,
+      "step": 6409
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.5028248587570624e-05,
+      "loss": 0.1627,
+      "step": 6410
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.502589453860641e-05,
+      "loss": 0.0684,
+      "step": 6411
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.5023540489642185e-05,
+      "loss": 0.1672,
+      "step": 6412
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.502118644067797e-05,
+      "loss": 0.1274,
+      "step": 6413
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.501883239171375e-05,
+      "loss": 0.2165,
+      "step": 6414
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.501647834274953e-05,
+      "loss": 0.1537,
+      "step": 6415
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.5014124293785314e-05,
+      "loss": 0.0827,
+      "step": 6416
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.50117702448211e-05,
+      "loss": 0.0639,
+      "step": 6417
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.5009416195856875e-05,
+      "loss": 0.0558,
+      "step": 6418
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.500706214689265e-05,
+      "loss": 0.1679,
+      "step": 6419
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.500470809792844e-05,
+      "loss": 0.1914,
+      "step": 6420
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.500235404896422e-05,
+      "loss": 0.1136,
+      "step": 6421
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.5e-05,
+      "loss": 0.1681,
+      "step": 6422
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.499764595103579e-05,
+      "loss": 0.1253,
+      "step": 6423
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4995291902071565e-05,
+      "loss": 0.0602,
+      "step": 6424
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.499293785310734e-05,
+      "loss": 0.0721,
+      "step": 6425
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4990583804143126e-05,
+      "loss": 0.088,
+      "step": 6426
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.498822975517891e-05,
+      "loss": 0.1345,
+      "step": 6427
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4985875706214694e-05,
+      "loss": 0.129,
+      "step": 6428
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.498352165725047e-05,
+      "loss": 0.1146,
+      "step": 6429
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4981167608286255e-05,
+      "loss": 0.1468,
+      "step": 6430
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.497881355932204e-05,
+      "loss": 0.1363,
+      "step": 6431
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4976459510357816e-05,
+      "loss": 0.2235,
+      "step": 6432
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.49741054613936e-05,
+      "loss": 0.1604,
+      "step": 6433
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4971751412429384e-05,
+      "loss": 0.0608,
+      "step": 6434
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.496939736346516e-05,
+      "loss": 0.1739,
+      "step": 6435
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.496704331450094e-05,
+      "loss": 0.0404,
+      "step": 6436
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.496468926553673e-05,
+      "loss": 0.0309,
+      "step": 6437
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4962335216572506e-05,
+      "loss": 0.1007,
+      "step": 6438
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.495998116760828e-05,
+      "loss": 0.1171,
+      "step": 6439
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4957627118644074e-05,
+      "loss": 0.1436,
+      "step": 6440
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.495527306967985e-05,
+      "loss": 0.1686,
+      "step": 6441
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.495291902071563e-05,
+      "loss": 0.1081,
+      "step": 6442
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.495056497175141e-05,
+      "loss": 0.1714,
+      "step": 6443
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4948210922787196e-05,
+      "loss": 0.0185,
+      "step": 6444
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.494585687382298e-05,
+      "loss": 0.069,
+      "step": 6445
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.494350282485876e-05,
+      "loss": 0.2249,
+      "step": 6446
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.494114877589454e-05,
+      "loss": 0.1326,
+      "step": 6447
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4938794726930325e-05,
+      "loss": 0.1118,
+      "step": 6448
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.49364406779661e-05,
+      "loss": 0.337,
+      "step": 6449
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4934086629001886e-05,
+      "loss": 0.0965,
+      "step": 6450
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.493173258003767e-05,
+      "loss": 0.2733,
+      "step": 6451
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.492937853107345e-05,
+      "loss": 0.1766,
+      "step": 6452
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4927024482109224e-05,
+      "loss": 0.0822,
+      "step": 6453
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4924670433145015e-05,
+      "loss": 0.0946,
+      "step": 6454
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.492231638418079e-05,
+      "loss": 0.1081,
+      "step": 6455
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.491996233521657e-05,
+      "loss": 0.2243,
+      "step": 6456
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.491760828625236e-05,
+      "loss": 0.1886,
+      "step": 6457
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.491525423728814e-05,
+      "loss": 0.1187,
+      "step": 6458
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.491290018832392e-05,
+      "loss": 0.1986,
+      "step": 6459
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.49105461393597e-05,
+      "loss": 0.2247,
+      "step": 6460
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.490819209039548e-05,
+      "loss": 0.1566,
+      "step": 6461
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4905838041431266e-05,
+      "loss": 0.0588,
+      "step": 6462
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.490348399246704e-05,
+      "loss": 0.1906,
+      "step": 6463
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.490112994350283e-05,
+      "loss": 0.1372,
+      "step": 6464
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.489877589453861e-05,
+      "loss": 0.1878,
+      "step": 6465
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.489642184557439e-05,
+      "loss": 0.253,
+      "step": 6466
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.489406779661017e-05,
+      "loss": 0.2542,
+      "step": 6467
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4891713747645956e-05,
+      "loss": 0.1224,
+      "step": 6468
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.488935969868173e-05,
+      "loss": 0.1766,
+      "step": 6469
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.488700564971752e-05,
+      "loss": 0.1317,
+      "step": 6470
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.48846516007533e-05,
+      "loss": 0.0432,
+      "step": 6471
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.488229755178908e-05,
+      "loss": 0.1335,
+      "step": 6472
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4879943502824855e-05,
+      "loss": 0.1094,
+      "step": 6473
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4877589453860646e-05,
+      "loss": 0.1846,
+      "step": 6474
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.487523540489642e-05,
+      "loss": 0.0535,
+      "step": 6475
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.487288135593221e-05,
+      "loss": 0.3115,
+      "step": 6476
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4870527306967984e-05,
+      "loss": 0.1796,
+      "step": 6477
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.486817325800377e-05,
+      "loss": 0.1432,
+      "step": 6478
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.486581920903955e-05,
+      "loss": 0.2297,
+      "step": 6479
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.486346516007533e-05,
+      "loss": 0.238,
+      "step": 6480
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.486111111111111e-05,
+      "loss": 0.1551,
+      "step": 6481
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.48587570621469e-05,
+      "loss": 0.0856,
+      "step": 6482
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4856403013182674e-05,
+      "loss": 0.1346,
+      "step": 6483
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.485404896421846e-05,
+      "loss": 0.1446,
+      "step": 6484
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.485169491525424e-05,
+      "loss": 0.1048,
+      "step": 6485
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.484934086629002e-05,
+      "loss": 0.1598,
+      "step": 6486
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.48469868173258e-05,
+      "loss": 0.0708,
+      "step": 6487
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.484463276836159e-05,
+      "loss": 0.1693,
+      "step": 6488
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4842278719397364e-05,
+      "loss": 0.2595,
+      "step": 6489
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.483992467043315e-05,
+      "loss": 0.1511,
+      "step": 6490
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.483757062146893e-05,
+      "loss": 0.1618,
+      "step": 6491
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.483521657250471e-05,
+      "loss": 0.0777,
+      "step": 6492
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.483286252354049e-05,
+      "loss": 0.0808,
+      "step": 6493
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.483050847457627e-05,
+      "loss": 0.1584,
+      "step": 6494
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4828154425612054e-05,
+      "loss": 0.1044,
+      "step": 6495
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.482580037664784e-05,
+      "loss": 0.0868,
+      "step": 6496
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4823446327683615e-05,
+      "loss": 0.1758,
+      "step": 6497
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.48210922787194e-05,
+      "loss": 0.1376,
+      "step": 6498
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.481873822975518e-05,
+      "loss": 0.2027,
+      "step": 6499
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.481638418079096e-05,
+      "loss": 0.1192,
+      "step": 6500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4814030131826744e-05,
+      "loss": 0.1893,
+      "step": 6501
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.481167608286253e-05,
+      "loss": 0.0915,
+      "step": 6502
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4809322033898305e-05,
+      "loss": 0.0956,
+      "step": 6503
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.480696798493409e-05,
+      "loss": 0.0874,
+      "step": 6504
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.480461393596987e-05,
+      "loss": 0.1937,
+      "step": 6505
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.480225988700565e-05,
+      "loss": 0.1053,
+      "step": 6506
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4799905838041434e-05,
+      "loss": 0.0772,
+      "step": 6507
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.479755178907722e-05,
+      "loss": 0.1827,
+      "step": 6508
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4795197740112995e-05,
+      "loss": 0.1643,
+      "step": 6509
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.479284369114878e-05,
+      "loss": 0.0523,
+      "step": 6510
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4790489642184556e-05,
+      "loss": 0.1618,
+      "step": 6511
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.478813559322034e-05,
+      "loss": 0.1895,
+      "step": 6512
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4785781544256124e-05,
+      "loss": 0.0832,
+      "step": 6513
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.47834274952919e-05,
+      "loss": 0.0641,
+      "step": 6514
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4781073446327685e-05,
+      "loss": 0.0839,
+      "step": 6515
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.477871939736347e-05,
+      "loss": 0.096,
+      "step": 6516
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4776365348399246e-05,
+      "loss": 0.2888,
+      "step": 6517
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.477401129943503e-05,
+      "loss": 0.1819,
+      "step": 6518
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4771657250470814e-05,
+      "loss": 0.1137,
+      "step": 6519
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.476930320150659e-05,
+      "loss": 0.1263,
+      "step": 6520
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4766949152542375e-05,
+      "loss": 0.0947,
+      "step": 6521
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.476459510357816e-05,
+      "loss": 0.061,
+      "step": 6522
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4762241054613936e-05,
+      "loss": 0.2198,
+      "step": 6523
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.475988700564972e-05,
+      "loss": 0.1012,
+      "step": 6524
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4757532956685504e-05,
+      "loss": 0.1389,
+      "step": 6525
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.475517890772128e-05,
+      "loss": 0.1304,
+      "step": 6526
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4752824858757065e-05,
+      "loss": 0.1033,
+      "step": 6527
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.475047080979284e-05,
+      "loss": 0.1507,
+      "step": 6528
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4748116760828626e-05,
+      "loss": 0.1341,
+      "step": 6529
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.474576271186441e-05,
+      "loss": 0.1308,
+      "step": 6530
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.474340866290019e-05,
+      "loss": 0.0549,
+      "step": 6531
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.474105461393597e-05,
+      "loss": 0.2341,
+      "step": 6532
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4738700564971755e-05,
+      "loss": 0.1247,
+      "step": 6533
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.473634651600753e-05,
+      "loss": 0.301,
+      "step": 6534
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4733992467043316e-05,
+      "loss": 0.0738,
+      "step": 6535
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.47316384180791e-05,
+      "loss": 0.1261,
+      "step": 6536
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.472928436911488e-05,
+      "loss": 0.1216,
+      "step": 6537
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.472693032015066e-05,
+      "loss": 0.1163,
+      "step": 6538
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4724576271186445e-05,
+      "loss": 0.1254,
+      "step": 6539
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.472222222222222e-05,
+      "loss": 0.1564,
+      "step": 6540
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4719868173258006e-05,
+      "loss": 0.1274,
+      "step": 6541
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.471751412429379e-05,
+      "loss": 0.1392,
+      "step": 6542
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.471516007532957e-05,
+      "loss": 0.2048,
+      "step": 6543
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.471280602636535e-05,
+      "loss": 0.1735,
+      "step": 6544
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.471045197740113e-05,
+      "loss": 0.0895,
+      "step": 6545
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.470809792843691e-05,
+      "loss": 0.071,
+      "step": 6546
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4705743879472696e-05,
+      "loss": 0.0817,
+      "step": 6547
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.470338983050847e-05,
+      "loss": 0.1677,
+      "step": 6548
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.470103578154426e-05,
+      "loss": 0.0787,
+      "step": 6549
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.469868173258004e-05,
+      "loss": 0.2504,
+      "step": 6550
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.469632768361582e-05,
+      "loss": 0.2109,
+      "step": 6551
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.46939736346516e-05,
+      "loss": 0.1677,
+      "step": 6552
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4691619585687386e-05,
+      "loss": 0.1257,
+      "step": 6553
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.468926553672316e-05,
+      "loss": 0.1075,
+      "step": 6554
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.468691148775895e-05,
+      "loss": 0.0943,
+      "step": 6555
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.468455743879473e-05,
+      "loss": 0.2437,
+      "step": 6556
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.468220338983051e-05,
+      "loss": 0.1333,
+      "step": 6557
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.467984934086629e-05,
+      "loss": 0.1333,
+      "step": 6558
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4677495291902076e-05,
+      "loss": 0.1347,
+      "step": 6559
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.467514124293785e-05,
+      "loss": 0.203,
+      "step": 6560
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.467278719397364e-05,
+      "loss": 0.0769,
+      "step": 6561
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.467043314500942e-05,
+      "loss": 0.1817,
+      "step": 6562
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.46680790960452e-05,
+      "loss": 0.0544,
+      "step": 6563
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.466572504708098e-05,
+      "loss": 0.1264,
+      "step": 6564
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.466337099811676e-05,
+      "loss": 0.1188,
+      "step": 6565
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.466101694915254e-05,
+      "loss": 0.0442,
+      "step": 6566
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.465866290018833e-05,
+      "loss": 0.2176,
+      "step": 6567
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4656308851224104e-05,
+      "loss": 0.0272,
+      "step": 6568
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.465395480225989e-05,
+      "loss": 0.1582,
+      "step": 6569
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.465160075329567e-05,
+      "loss": 0.1984,
+      "step": 6570
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.464924670433145e-05,
+      "loss": 0.141,
+      "step": 6571
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.464689265536723e-05,
+      "loss": 0.0325,
+      "step": 6572
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.464453860640302e-05,
+      "loss": 0.069,
+      "step": 6573
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4642184557438794e-05,
+      "loss": 0.1953,
+      "step": 6574
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.463983050847458e-05,
+      "loss": 0.1761,
+      "step": 6575
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.463747645951036e-05,
+      "loss": 0.312,
+      "step": 6576
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.463512241054614e-05,
+      "loss": 0.119,
+      "step": 6577
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.463276836158192e-05,
+      "loss": 0.0943,
+      "step": 6578
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.463041431261771e-05,
+      "loss": 0.1099,
+      "step": 6579
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4628060263653484e-05,
+      "loss": 0.1784,
+      "step": 6580
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.462570621468927e-05,
+      "loss": 0.0667,
+      "step": 6581
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4623352165725045e-05,
+      "loss": 0.1342,
+      "step": 6582
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.462099811676083e-05,
+      "loss": 0.258,
+      "step": 6583
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.461864406779661e-05,
+      "loss": 0.0656,
+      "step": 6584
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.461629001883239e-05,
+      "loss": 0.0673,
+      "step": 6585
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4613935969868174e-05,
+      "loss": 0.1294,
+      "step": 6586
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.461158192090396e-05,
+      "loss": 0.1312,
+      "step": 6587
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4609227871939735e-05,
+      "loss": 0.1307,
+      "step": 6588
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.460687382297552e-05,
+      "loss": 0.163,
+      "step": 6589
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.46045197740113e-05,
+      "loss": 0.0845,
+      "step": 6590
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.460216572504708e-05,
+      "loss": 0.1256,
+      "step": 6591
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4599811676082864e-05,
+      "loss": 0.162,
+      "step": 6592
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.459745762711865e-05,
+      "loss": 0.0823,
+      "step": 6593
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4595103578154425e-05,
+      "loss": 0.2441,
+      "step": 6594
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.459274952919021e-05,
+      "loss": 0.1632,
+      "step": 6595
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.459039548022599e-05,
+      "loss": 0.2997,
+      "step": 6596
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.458804143126177e-05,
+      "loss": 0.172,
+      "step": 6597
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4585687382297554e-05,
+      "loss": 0.1283,
+      "step": 6598
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.458333333333333e-05,
+      "loss": 0.0552,
+      "step": 6599
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4580979284369115e-05,
+      "loss": 0.1988,
+      "step": 6600
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.45786252354049e-05,
+      "loss": 0.1148,
+      "step": 6601
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4576271186440676e-05,
+      "loss": 0.1728,
+      "step": 6602
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.457391713747646e-05,
+      "loss": 0.1594,
+      "step": 6603
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4571563088512244e-05,
+      "loss": 0.0747,
+      "step": 6604
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.456920903954802e-05,
+      "loss": 0.0926,
+      "step": 6605
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4566854990583805e-05,
+      "loss": 0.1488,
+      "step": 6606
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.456450094161959e-05,
+      "loss": 0.0829,
+      "step": 6607
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4562146892655366e-05,
+      "loss": 0.139,
+      "step": 6608
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.455979284369115e-05,
+      "loss": 0.1501,
+      "step": 6609
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4557438794726934e-05,
+      "loss": 0.1908,
+      "step": 6610
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.455508474576271e-05,
+      "loss": 0.132,
+      "step": 6611
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4552730696798495e-05,
+      "loss": 0.1676,
+      "step": 6612
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.455037664783428e-05,
+      "loss": 0.114,
+      "step": 6613
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4548022598870056e-05,
+      "loss": 0.1363,
+      "step": 6614
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.454566854990584e-05,
+      "loss": 0.1966,
+      "step": 6615
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.454331450094162e-05,
+      "loss": 0.1051,
+      "step": 6616
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.454096045197741e-05,
+      "loss": 0.108,
+      "step": 6617
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4538606403013185e-05,
+      "loss": 0.1909,
+      "step": 6618
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.453625235404896e-05,
+      "loss": 0.2684,
+      "step": 6619
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4533898305084746e-05,
+      "loss": 0.1098,
+      "step": 6620
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.453154425612053e-05,
+      "loss": 0.0708,
+      "step": 6621
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.452919020715631e-05,
+      "loss": 0.1131,
+      "step": 6622
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.452683615819209e-05,
+      "loss": 0.1281,
+      "step": 6623
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4524482109227875e-05,
+      "loss": 0.2001,
+      "step": 6624
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.452212806026365e-05,
+      "loss": 0.1942,
+      "step": 6625
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4519774011299436e-05,
+      "loss": 0.2921,
+      "step": 6626
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.451741996233522e-05,
+      "loss": 0.1183,
+      "step": 6627
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4515065913371e-05,
+      "loss": 0.1585,
+      "step": 6628
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.451271186440678e-05,
+      "loss": 0.1002,
+      "step": 6629
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4510357815442565e-05,
+      "loss": 0.1961,
+      "step": 6630
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.450800376647834e-05,
+      "loss": 0.1645,
+      "step": 6631
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4505649717514126e-05,
+      "loss": 0.1288,
+      "step": 6632
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4503295668549904e-05,
+      "loss": 0.1797,
+      "step": 6633
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4500941619585694e-05,
+      "loss": 0.0714,
+      "step": 6634
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.449858757062147e-05,
+      "loss": 0.115,
+      "step": 6635
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.449623352165725e-05,
+      "loss": 0.1678,
+      "step": 6636
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.449387947269304e-05,
+      "loss": 0.1487,
+      "step": 6637
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4491525423728816e-05,
+      "loss": 0.1724,
+      "step": 6638
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4489171374764594e-05,
+      "loss": 0.1614,
+      "step": 6639
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.448681732580038e-05,
+      "loss": 0.1892,
+      "step": 6640
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.448446327683616e-05,
+      "loss": 0.0941,
+      "step": 6641
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.448210922787194e-05,
+      "loss": 0.1252,
+      "step": 6642
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.447975517890772e-05,
+      "loss": 0.2056,
+      "step": 6643
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4477401129943506e-05,
+      "loss": 0.1345,
+      "step": 6644
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4475047080979284e-05,
+      "loss": 0.1087,
+      "step": 6645
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.447269303201507e-05,
+      "loss": 0.1303,
+      "step": 6646
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.447033898305085e-05,
+      "loss": 0.1744,
+      "step": 6647
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4467984934086635e-05,
+      "loss": 0.1392,
+      "step": 6648
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.446563088512241e-05,
+      "loss": 0.0821,
+      "step": 6649
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.446327683615819e-05,
+      "loss": 0.0686,
+      "step": 6650
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.446092278719398e-05,
+      "loss": 0.097,
+      "step": 6651
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.445856873822976e-05,
+      "loss": 0.0901,
+      "step": 6652
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4456214689265535e-05,
+      "loss": 0.2539,
+      "step": 6653
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4453860640301325e-05,
+      "loss": 0.256,
+      "step": 6654
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.44515065913371e-05,
+      "loss": 0.1698,
+      "step": 6655
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.444915254237288e-05,
+      "loss": 0.1356,
+      "step": 6656
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4446798493408664e-05,
+      "loss": 0.0834,
+      "step": 6657
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.444444444444445e-05,
+      "loss": 0.1404,
+      "step": 6658
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4442090395480225e-05,
+      "loss": 0.1466,
+      "step": 6659
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.443973634651601e-05,
+      "loss": 0.0647,
+      "step": 6660
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.443738229755179e-05,
+      "loss": 0.1483,
+      "step": 6661
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.443502824858757e-05,
+      "loss": 0.1536,
+      "step": 6662
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4432674199623354e-05,
+      "loss": 0.1083,
+      "step": 6663
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.443032015065914e-05,
+      "loss": 0.1399,
+      "step": 6664
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.442796610169492e-05,
+      "loss": 0.1586,
+      "step": 6665
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.44256120527307e-05,
+      "loss": 0.0778,
+      "step": 6666
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4423258003766476e-05,
+      "loss": 0.3139,
+      "step": 6667
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4420903954802266e-05,
+      "loss": 0.1155,
+      "step": 6668
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4418549905838044e-05,
+      "loss": 0.179,
+      "step": 6669
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.441619585687382e-05,
+      "loss": 0.1483,
+      "step": 6670
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.441384180790961e-05,
+      "loss": 0.1721,
+      "step": 6671
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.441148775894539e-05,
+      "loss": 0.0717,
+      "step": 6672
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4409133709981166e-05,
+      "loss": 0.1862,
+      "step": 6673
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.440677966101695e-05,
+      "loss": 0.1613,
+      "step": 6674
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4404425612052734e-05,
+      "loss": 0.0928,
+      "step": 6675
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.440207156308851e-05,
+      "loss": 0.1681,
+      "step": 6676
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4399717514124295e-05,
+      "loss": 0.117,
+      "step": 6677
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.439736346516008e-05,
+      "loss": 0.152,
+      "step": 6678
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4395009416195856e-05,
+      "loss": 0.1124,
+      "step": 6679
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.439265536723164e-05,
+      "loss": 0.0754,
+      "step": 6680
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4390301318267424e-05,
+      "loss": 0.162,
+      "step": 6681
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.438794726930321e-05,
+      "loss": 0.1074,
+      "step": 6682
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4385593220338985e-05,
+      "loss": 0.1514,
+      "step": 6683
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.438323917137476e-05,
+      "loss": 0.056,
+      "step": 6684
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.438088512241055e-05,
+      "loss": 0.0665,
+      "step": 6685
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.437853107344633e-05,
+      "loss": 0.1412,
+      "step": 6686
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.437617702448211e-05,
+      "loss": 0.1926,
+      "step": 6687
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.43738229755179e-05,
+      "loss": 0.1174,
+      "step": 6688
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4371468926553675e-05,
+      "loss": 0.0763,
+      "step": 6689
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.436911487758945e-05,
+      "loss": 0.0619,
+      "step": 6690
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4366760828625236e-05,
+      "loss": 0.2432,
+      "step": 6691
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.436440677966102e-05,
+      "loss": 0.1475,
+      "step": 6692
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.43620527306968e-05,
+      "loss": 0.1934,
+      "step": 6693
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.435969868173258e-05,
+      "loss": 0.1992,
+      "step": 6694
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4357344632768365e-05,
+      "loss": 0.1364,
+      "step": 6695
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.435499058380415e-05,
+      "loss": 0.1819,
+      "step": 6696
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4352636534839926e-05,
+      "loss": 0.1461,
+      "step": 6697
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.435028248587571e-05,
+      "loss": 0.2673,
+      "step": 6698
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4347928436911493e-05,
+      "loss": 0.2053,
+      "step": 6699
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.434557438794727e-05,
+      "loss": 0.1912,
+      "step": 6700
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.434322033898305e-05,
+      "loss": 0.0709,
+      "step": 6701
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.434086629001884e-05,
+      "loss": 0.124,
+      "step": 6702
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4338512241054616e-05,
+      "loss": 0.0797,
+      "step": 6703
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.433615819209039e-05,
+      "loss": 0.1943,
+      "step": 6704
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4333804143126183e-05,
+      "loss": 0.1159,
+      "step": 6705
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.433145009416196e-05,
+      "loss": 0.0768,
+      "step": 6706
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.432909604519774e-05,
+      "loss": 0.0702,
+      "step": 6707
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.432674199623352e-05,
+      "loss": 0.0989,
+      "step": 6708
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4324387947269306e-05,
+      "loss": 0.1328,
+      "step": 6709
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.432203389830508e-05,
+      "loss": 0.0551,
+      "step": 6710
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.431967984934087e-05,
+      "loss": 0.1097,
+      "step": 6711
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.431732580037665e-05,
+      "loss": 0.0724,
+      "step": 6712
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4314971751412435e-05,
+      "loss": 0.1837,
+      "step": 6713
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.431261770244821e-05,
+      "loss": 0.2367,
+      "step": 6714
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4310263653483996e-05,
+      "loss": 0.1028,
+      "step": 6715
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.430790960451978e-05,
+      "loss": 0.2034,
+      "step": 6716
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.430555555555556e-05,
+      "loss": 0.077,
+      "step": 6717
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4303201506591334e-05,
+      "loss": 0.1673,
+      "step": 6718
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4300847457627125e-05,
+      "loss": 0.2007,
+      "step": 6719
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.42984934086629e-05,
+      "loss": 0.0914,
+      "step": 6720
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.429613935969868e-05,
+      "loss": 0.2013,
+      "step": 6721
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.429378531073447e-05,
+      "loss": 0.078,
+      "step": 6722
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.429143126177025e-05,
+      "loss": 0.2196,
+      "step": 6723
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4289077212806024e-05,
+      "loss": 0.2236,
+      "step": 6724
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.428672316384181e-05,
+      "loss": 0.2011,
+      "step": 6725
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.428436911487759e-05,
+      "loss": 0.1162,
+      "step": 6726
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.428201506591337e-05,
+      "loss": 0.0953,
+      "step": 6727
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.427966101694915e-05,
+      "loss": 0.3142,
+      "step": 6728
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.427730696798494e-05,
+      "loss": 0.028,
+      "step": 6729
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.427495291902072e-05,
+      "loss": 0.3174,
+      "step": 6730
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.42725988700565e-05,
+      "loss": 0.0902,
+      "step": 6731
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.427024482109228e-05,
+      "loss": 0.0693,
+      "step": 6732
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4267890772128066e-05,
+      "loss": 0.0859,
+      "step": 6733
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.426553672316384e-05,
+      "loss": 0.1469,
+      "step": 6734
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.426318267419962e-05,
+      "loss": 0.0426,
+      "step": 6735
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.426082862523541e-05,
+      "loss": 0.2694,
+      "step": 6736
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.425847457627119e-05,
+      "loss": 0.111,
+      "step": 6737
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4256120527306965e-05,
+      "loss": 0.0559,
+      "step": 6738
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4253766478342756e-05,
+      "loss": 0.1858,
+      "step": 6739
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.425141242937853e-05,
+      "loss": 0.1092,
+      "step": 6740
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.424905838041431e-05,
+      "loss": 0.0987,
+      "step": 6741
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4246704331450094e-05,
+      "loss": 0.0614,
+      "step": 6742
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.424435028248588e-05,
+      "loss": 0.1193,
+      "step": 6743
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.424199623352166e-05,
+      "loss": 0.0298,
+      "step": 6744
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.423964218455744e-05,
+      "loss": 0.1337,
+      "step": 6745
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.423728813559322e-05,
+      "loss": 0.1947,
+      "step": 6746
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.423493408662901e-05,
+      "loss": 0.0402,
+      "step": 6747
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4232580037664784e-05,
+      "loss": 0.0698,
+      "step": 6748
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.423022598870057e-05,
+      "loss": 0.2576,
+      "step": 6749
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.422787193973635e-05,
+      "loss": 0.1451,
+      "step": 6750
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.422551789077213e-05,
+      "loss": 0.1371,
+      "step": 6751
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4223163841807906e-05,
+      "loss": 0.1319,
+      "step": 6752
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.42208097928437e-05,
+      "loss": 0.2152,
+      "step": 6753
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4218455743879474e-05,
+      "loss": 0.1154,
+      "step": 6754
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.421610169491525e-05,
+      "loss": 0.0676,
+      "step": 6755
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.421374764595104e-05,
+      "loss": 0.0721,
+      "step": 6756
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.421139359698682e-05,
+      "loss": 0.0955,
+      "step": 6757
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4209039548022596e-05,
+      "loss": 0.1664,
+      "step": 6758
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.420668549905838e-05,
+      "loss": 0.1802,
+      "step": 6759
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4204331450094164e-05,
+      "loss": 0.2009,
+      "step": 6760
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.420197740112995e-05,
+      "loss": 0.1944,
+      "step": 6761
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4199623352165725e-05,
+      "loss": 0.1226,
+      "step": 6762
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.419726930320151e-05,
+      "loss": 0.2678,
+      "step": 6763
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.419491525423729e-05,
+      "loss": 0.2885,
+      "step": 6764
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.419256120527307e-05,
+      "loss": 0.044,
+      "step": 6765
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4190207156308854e-05,
+      "loss": 0.0994,
+      "step": 6766
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.418785310734464e-05,
+      "loss": 0.178,
+      "step": 6767
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4185499058380415e-05,
+      "loss": 0.0539,
+      "step": 6768
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.418314500941619e-05,
+      "loss": 0.103,
+      "step": 6769
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.418079096045198e-05,
+      "loss": 0.0502,
+      "step": 6770
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.417843691148776e-05,
+      "loss": 0.0622,
+      "step": 6771
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.417608286252354e-05,
+      "loss": 0.2217,
+      "step": 6772
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.417372881355933e-05,
+      "loss": 0.1577,
+      "step": 6773
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4171374764595105e-05,
+      "loss": 0.1709,
+      "step": 6774
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.416902071563089e-05,
+      "loss": 0.2146,
+      "step": 6775
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4166666666666666e-05,
+      "loss": 0.298,
+      "step": 6776
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.416431261770245e-05,
+      "loss": 0.1346,
+      "step": 6777
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4161958568738234e-05,
+      "loss": 0.2076,
+      "step": 6778
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.415960451977401e-05,
+      "loss": 0.1307,
+      "step": 6779
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4157250470809795e-05,
+      "loss": 0.1617,
+      "step": 6780
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.415489642184558e-05,
+      "loss": 0.0569,
+      "step": 6781
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4152542372881356e-05,
+      "loss": 0.2854,
+      "step": 6782
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.415018832391714e-05,
+      "loss": 0.1301,
+      "step": 6783
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4147834274952924e-05,
+      "loss": 0.1094,
+      "step": 6784
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.41454802259887e-05,
+      "loss": 0.1469,
+      "step": 6785
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4143126177024485e-05,
+      "loss": 0.1847,
+      "step": 6786
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.414077212806027e-05,
+      "loss": 0.0642,
+      "step": 6787
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4138418079096046e-05,
+      "loss": 0.2338,
+      "step": 6788
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.413606403013182e-05,
+      "loss": 0.0733,
+      "step": 6789
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4133709981167614e-05,
+      "loss": 0.1221,
+      "step": 6790
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.413135593220339e-05,
+      "loss": 0.116,
+      "step": 6791
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4129001883239175e-05,
+      "loss": 0.0826,
+      "step": 6792
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.412664783427495e-05,
+      "loss": 0.1378,
+      "step": 6793
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4124293785310736e-05,
+      "loss": 0.1293,
+      "step": 6794
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.412193973634652e-05,
+      "loss": 0.1452,
+      "step": 6795
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.41195856873823e-05,
+      "loss": 0.1155,
+      "step": 6796
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.411723163841808e-05,
+      "loss": 0.1625,
+      "step": 6797
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4114877589453865e-05,
+      "loss": 0.1074,
+      "step": 6798
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.411252354048964e-05,
+      "loss": 0.0777,
+      "step": 6799
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4110169491525426e-05,
+      "loss": 0.0599,
+      "step": 6800
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.410781544256121e-05,
+      "loss": 0.1227,
+      "step": 6801
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.410546139359699e-05,
+      "loss": 0.1701,
+      "step": 6802
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.410310734463277e-05,
+      "loss": 0.2001,
+      "step": 6803
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4100753295668555e-05,
+      "loss": 0.0591,
+      "step": 6804
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.409839924670433e-05,
+      "loss": 0.2045,
+      "step": 6805
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.409604519774011e-05,
+      "loss": 0.2402,
+      "step": 6806
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.40936911487759e-05,
+      "loss": 0.3355,
+      "step": 6807
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.409133709981168e-05,
+      "loss": 0.0866,
+      "step": 6808
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.408898305084746e-05,
+      "loss": 0.0426,
+      "step": 6809
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.408662900188324e-05,
+      "loss": 0.0528,
+      "step": 6810
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.408427495291902e-05,
+      "loss": 0.1294,
+      "step": 6811
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4081920903954806e-05,
+      "loss": 0.2635,
+      "step": 6812
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.407956685499058e-05,
+      "loss": 0.0952,
+      "step": 6813
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.407721280602637e-05,
+      "loss": 0.1579,
+      "step": 6814
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.407485875706215e-05,
+      "loss": 0.1902,
+      "step": 6815
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.407250470809793e-05,
+      "loss": 0.1161,
+      "step": 6816
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.407015065913371e-05,
+      "loss": 0.0588,
+      "step": 6817
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4067796610169496e-05,
+      "loss": 0.2016,
+      "step": 6818
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.406544256120527e-05,
+      "loss": 0.1452,
+      "step": 6819
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.406308851224106e-05,
+      "loss": 0.2205,
+      "step": 6820
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.406073446327684e-05,
+      "loss": 0.1192,
+      "step": 6821
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.405838041431262e-05,
+      "loss": 0.1074,
+      "step": 6822
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.40560263653484e-05,
+      "loss": 0.1642,
+      "step": 6823
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.4053672316384186e-05,
+      "loss": 0.1457,
+      "step": 6824
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.405131826741996e-05,
+      "loss": 0.0625,
+      "step": 6825
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.404896421845575e-05,
+      "loss": 0.1025,
+      "step": 6826
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.4046610169491524e-05,
+      "loss": 0.0962,
+      "step": 6827
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.404425612052731e-05,
+      "loss": 0.1288,
+      "step": 6828
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.404190207156309e-05,
+      "loss": 0.1568,
+      "step": 6829
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.403954802259887e-05,
+      "loss": 0.1921,
+      "step": 6830
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.403719397363465e-05,
+      "loss": 0.1334,
+      "step": 6831
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.403483992467044e-05,
+      "loss": 0.1922,
+      "step": 6832
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.4032485875706214e-05,
+      "loss": 0.1185,
+      "step": 6833
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.4030131826742e-05,
+      "loss": 0.1665,
+      "step": 6834
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.402777777777778e-05,
+      "loss": 0.1006,
+      "step": 6835
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.402542372881356e-05,
+      "loss": 0.0966,
+      "step": 6836
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.402306967984934e-05,
+      "loss": 0.13,
+      "step": 6837
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.402071563088513e-05,
+      "loss": 0.2193,
+      "step": 6838
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.4018361581920904e-05,
+      "loss": 0.0895,
+      "step": 6839
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.401600753295669e-05,
+      "loss": 0.1797,
+      "step": 6840
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.401365348399247e-05,
+      "loss": 0.2481,
+      "step": 6841
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.401129943502825e-05,
+      "loss": 0.1105,
+      "step": 6842
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.400894538606403e-05,
+      "loss": 0.1432,
+      "step": 6843
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.400659133709981e-05,
+      "loss": 0.1583,
+      "step": 6844
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.4004237288135594e-05,
+      "loss": 0.0483,
+      "step": 6845
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.400188323917138e-05,
+      "loss": 0.1856,
+      "step": 6846
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3999529190207155e-05,
+      "loss": 0.0714,
+      "step": 6847
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.399717514124294e-05,
+      "loss": 0.1871,
+      "step": 6848
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.399482109227872e-05,
+      "loss": 0.1043,
+      "step": 6849
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.39924670433145e-05,
+      "loss": 0.3134,
+      "step": 6850
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3990112994350284e-05,
+      "loss": 0.2954,
+      "step": 6851
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.398775894538607e-05,
+      "loss": 0.1031,
+      "step": 6852
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3985404896421845e-05,
+      "loss": 0.296,
+      "step": 6853
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.398305084745763e-05,
+      "loss": 0.0956,
+      "step": 6854
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.398069679849341e-05,
+      "loss": 0.1678,
+      "step": 6855
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.397834274952919e-05,
+      "loss": 0.0773,
+      "step": 6856
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3975988700564974e-05,
+      "loss": 0.0547,
+      "step": 6857
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.397363465160076e-05,
+      "loss": 0.2461,
+      "step": 6858
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3971280602636535e-05,
+      "loss": 0.0316,
+      "step": 6859
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.396892655367232e-05,
+      "loss": 0.1162,
+      "step": 6860
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3966572504708096e-05,
+      "loss": 0.1717,
+      "step": 6861
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.396421845574388e-05,
+      "loss": 0.1587,
+      "step": 6862
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3961864406779664e-05,
+      "loss": 0.0908,
+      "step": 6863
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.395951035781544e-05,
+      "loss": 0.0699,
+      "step": 6864
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3957156308851225e-05,
+      "loss": 0.2692,
+      "step": 6865
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.395480225988701e-05,
+      "loss": 0.0931,
+      "step": 6866
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3952448210922786e-05,
+      "loss": 0.2582,
+      "step": 6867
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.395009416195857e-05,
+      "loss": 0.1814,
+      "step": 6868
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3947740112994354e-05,
+      "loss": 0.0523,
+      "step": 6869
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.394538606403013e-05,
+      "loss": 0.0627,
+      "step": 6870
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3943032015065915e-05,
+      "loss": 0.1915,
+      "step": 6871
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.39406779661017e-05,
+      "loss": 0.1724,
+      "step": 6872
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3938323917137476e-05,
+      "loss": 0.0536,
+      "step": 6873
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.393596986817326e-05,
+      "loss": 0.212,
+      "step": 6874
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3933615819209044e-05,
+      "loss": 0.1004,
+      "step": 6875
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.393126177024482e-05,
+      "loss": 0.1845,
+      "step": 6876
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3928907721280605e-05,
+      "loss": 0.2218,
+      "step": 6877
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.392655367231639e-05,
+      "loss": 0.134,
+      "step": 6878
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3924199623352166e-05,
+      "loss": 0.1733,
+      "step": 6879
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.392184557438795e-05,
+      "loss": 0.11,
+      "step": 6880
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.391949152542373e-05,
+      "loss": 0.0704,
+      "step": 6881
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.391713747645951e-05,
+      "loss": 0.2194,
+      "step": 6882
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3914783427495295e-05,
+      "loss": 0.1202,
+      "step": 6883
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.391242937853107e-05,
+      "loss": 0.0757,
+      "step": 6884
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3910075329566856e-05,
+      "loss": 0.094,
+      "step": 6885
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.390772128060264e-05,
+      "loss": 0.2153,
+      "step": 6886
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.390536723163842e-05,
+      "loss": 0.1092,
+      "step": 6887
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.39030131826742e-05,
+      "loss": 0.0806,
+      "step": 6888
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3900659133709985e-05,
+      "loss": 0.1508,
+      "step": 6889
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.389830508474576e-05,
+      "loss": 0.1944,
+      "step": 6890
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3895951035781546e-05,
+      "loss": 0.1989,
+      "step": 6891
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.389359698681733e-05,
+      "loss": 0.1786,
+      "step": 6892
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.389124293785311e-05,
+      "loss": 0.0432,
+      "step": 6893
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.388888888888889e-05,
+      "loss": 0.1228,
+      "step": 6894
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3886534839924675e-05,
+      "loss": 0.2214,
+      "step": 6895
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.388418079096045e-05,
+      "loss": 0.1246,
+      "step": 6896
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3881826741996236e-05,
+      "loss": 0.1004,
+      "step": 6897
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.387947269303201e-05,
+      "loss": 0.1384,
+      "step": 6898
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.38771186440678e-05,
+      "loss": 0.2273,
+      "step": 6899
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.387476459510358e-05,
+      "loss": 0.1162,
+      "step": 6900
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.387241054613936e-05,
+      "loss": 0.1554,
+      "step": 6901
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.387005649717514e-05,
+      "loss": 0.1054,
+      "step": 6902
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3867702448210926e-05,
+      "loss": 0.1084,
+      "step": 6903
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.38653483992467e-05,
+      "loss": 0.0948,
+      "step": 6904
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.386299435028249e-05,
+      "loss": 0.1527,
+      "step": 6905
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.386064030131827e-05,
+      "loss": 0.1178,
+      "step": 6906
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.385828625235405e-05,
+      "loss": 0.0547,
+      "step": 6907
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.385593220338983e-05,
+      "loss": 0.1357,
+      "step": 6908
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3853578154425616e-05,
+      "loss": 0.1725,
+      "step": 6909
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.385122410546139e-05,
+      "loss": 0.0692,
+      "step": 6910
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.384887005649718e-05,
+      "loss": 0.1396,
+      "step": 6911
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.384651600753296e-05,
+      "loss": 0.2018,
+      "step": 6912
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.384416195856874e-05,
+      "loss": 0.0433,
+      "step": 6913
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.384180790960452e-05,
+      "loss": 0.0793,
+      "step": 6914
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.38394538606403e-05,
+      "loss": 0.1596,
+      "step": 6915
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.383709981167608e-05,
+      "loss": 0.1594,
+      "step": 6916
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.383474576271187e-05,
+      "loss": 0.1483,
+      "step": 6917
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3832391713747644e-05,
+      "loss": 0.2356,
+      "step": 6918
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.383003766478343e-05,
+      "loss": 0.0819,
+      "step": 6919
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.382768361581921e-05,
+      "loss": 0.1014,
+      "step": 6920
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.382532956685499e-05,
+      "loss": 0.1002,
+      "step": 6921
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.382297551789077e-05,
+      "loss": 0.1115,
+      "step": 6922
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.382062146892656e-05,
+      "loss": 0.0918,
+      "step": 6923
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3818267419962334e-05,
+      "loss": 0.1429,
+      "step": 6924
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.381591337099812e-05,
+      "loss": 0.2875,
+      "step": 6925
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.38135593220339e-05,
+      "loss": 0.1212,
+      "step": 6926
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.381120527306968e-05,
+      "loss": 0.2315,
+      "step": 6927
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.380885122410546e-05,
+      "loss": 0.2272,
+      "step": 6928
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.380649717514125e-05,
+      "loss": 0.142,
+      "step": 6929
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3804143126177024e-05,
+      "loss": 0.1896,
+      "step": 6930
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.380178907721281e-05,
+      "loss": 0.284,
+      "step": 6931
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3799435028248585e-05,
+      "loss": 0.1382,
+      "step": 6932
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3797080979284376e-05,
+      "loss": 0.0658,
+      "step": 6933
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.379472693032015e-05,
+      "loss": 0.0746,
+      "step": 6934
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.379237288135593e-05,
+      "loss": 0.2111,
+      "step": 6935
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3790018832391714e-05,
+      "loss": 0.2738,
+      "step": 6936
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.37876647834275e-05,
+      "loss": 0.1973,
+      "step": 6937
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3785310734463275e-05,
+      "loss": 0.191,
+      "step": 6938
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.378295668549906e-05,
+      "loss": 0.1249,
+      "step": 6939
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.378060263653484e-05,
+      "loss": 0.0994,
+      "step": 6940
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.377824858757062e-05,
+      "loss": 0.1025,
+      "step": 6941
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3775894538606404e-05,
+      "loss": 0.0973,
+      "step": 6942
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.377354048964219e-05,
+      "loss": 0.0778,
+      "step": 6943
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3771186440677965e-05,
+      "loss": 0.1266,
+      "step": 6944
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.376883239171375e-05,
+      "loss": 0.1885,
+      "step": 6945
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.376647834274953e-05,
+      "loss": 0.0362,
+      "step": 6946
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.376412429378531e-05,
+      "loss": 0.1161,
+      "step": 6947
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3761770244821094e-05,
+      "loss": 0.1293,
+      "step": 6948
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.375941619585687e-05,
+      "loss": 0.1076,
+      "step": 6949
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.375706214689266e-05,
+      "loss": 0.0871,
+      "step": 6950
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.375470809792844e-05,
+      "loss": 0.14,
+      "step": 6951
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3752354048964217e-05,
+      "loss": 0.0398,
+      "step": 6952
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.375000000000001e-05,
+      "loss": 0.1016,
+      "step": 6953
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3747645951035784e-05,
+      "loss": 0.0893,
+      "step": 6954
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.374529190207156e-05,
+      "loss": 0.1054,
+      "step": 6955
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3742937853107345e-05,
+      "loss": 0.1316,
+      "step": 6956
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.374058380414313e-05,
+      "loss": 0.11,
+      "step": 6957
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3738229755178907e-05,
+      "loss": 0.1863,
+      "step": 6958
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.373587570621469e-05,
+      "loss": 0.3213,
+      "step": 6959
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3733521657250474e-05,
+      "loss": 0.0734,
+      "step": 6960
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.373116760828625e-05,
+      "loss": 0.1152,
+      "step": 6961
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3728813559322035e-05,
+      "loss": 0.1379,
+      "step": 6962
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.372645951035782e-05,
+      "loss": 0.1385,
+      "step": 6963
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3724105461393596e-05,
+      "loss": 0.1023,
+      "step": 6964
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.372175141242938e-05,
+      "loss": 0.2728,
+      "step": 6965
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.371939736346516e-05,
+      "loss": 0.2371,
+      "step": 6966
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.371704331450095e-05,
+      "loss": 0.3579,
+      "step": 6967
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3714689265536725e-05,
+      "loss": 0.0898,
+      "step": 6968
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.37123352165725e-05,
+      "loss": 0.2109,
+      "step": 6969
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.370998116760829e-05,
+      "loss": 0.1574,
+      "step": 6970
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.370762711864407e-05,
+      "loss": 0.2117,
+      "step": 6971
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.370527306967985e-05,
+      "loss": 0.1025,
+      "step": 6972
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.370291902071563e-05,
+      "loss": 0.2416,
+      "step": 6973
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3700564971751415e-05,
+      "loss": 0.1363,
+      "step": 6974
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.369821092278719e-05,
+      "loss": 0.2549,
+      "step": 6975
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3695856873822976e-05,
+      "loss": 0.0737,
+      "step": 6976
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.369350282485876e-05,
+      "loss": 0.1269,
+      "step": 6977
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.369114877589454e-05,
+      "loss": 0.2092,
+      "step": 6978
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.368879472693032e-05,
+      "loss": 0.1361,
+      "step": 6979
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3686440677966105e-05,
+      "loss": 0.1903,
+      "step": 6980
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.368408662900189e-05,
+      "loss": 0.1445,
+      "step": 6981
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3681732580037666e-05,
+      "loss": 0.103,
+      "step": 6982
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3679378531073444e-05,
+      "loss": 0.2177,
+      "step": 6983
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3677024482109234e-05,
+      "loss": 0.2815,
+      "step": 6984
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.367467043314501e-05,
+      "loss": 0.1443,
+      "step": 6985
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.367231638418079e-05,
+      "loss": 0.1453,
+      "step": 6986
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.366996233521658e-05,
+      "loss": 0.1225,
+      "step": 6987
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3667608286252356e-05,
+      "loss": 0.2268,
+      "step": 6988
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3665254237288134e-05,
+      "loss": 0.0906,
+      "step": 6989
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.366290018832392e-05,
+      "loss": 0.0459,
+      "step": 6990
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.36605461393597e-05,
+      "loss": 0.143,
+      "step": 6991
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.365819209039548e-05,
+      "loss": 0.2001,
+      "step": 6992
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.365583804143126e-05,
+      "loss": 0.1527,
+      "step": 6993
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3653483992467046e-05,
+      "loss": 0.0952,
+      "step": 6994
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3651129943502824e-05,
+      "loss": 0.1353,
+      "step": 6995
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.364877589453861e-05,
+      "loss": 0.1208,
+      "step": 6996
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.364642184557439e-05,
+      "loss": 0.2371,
+      "step": 6997
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3644067796610175e-05,
+      "loss": 0.0363,
+      "step": 6998
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.364171374764595e-05,
+      "loss": 0.1407,
+      "step": 6999
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.363935969868173e-05,
+      "loss": 0.1221,
+      "step": 7000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.363700564971752e-05,
+      "loss": 0.107,
+      "step": 7001
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.36346516007533e-05,
+      "loss": 0.1832,
+      "step": 7002
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3632297551789075e-05,
+      "loss": 0.1622,
+      "step": 7003
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3629943502824865e-05,
+      "loss": 0.1218,
+      "step": 7004
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.362758945386064e-05,
+      "loss": 0.2505,
+      "step": 7005
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.362523540489642e-05,
+      "loss": 0.2654,
+      "step": 7006
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3622881355932204e-05,
+      "loss": 0.1228,
+      "step": 7007
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.362052730696799e-05,
+      "loss": 0.1415,
+      "step": 7008
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3618173258003765e-05,
+      "loss": 0.0737,
+      "step": 7009
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.361581920903955e-05,
+      "loss": 0.1904,
+      "step": 7010
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.361346516007533e-05,
+      "loss": 0.1976,
+      "step": 7011
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3611111111111116e-05,
+      "loss": 0.0981,
+      "step": 7012
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3608757062146894e-05,
+      "loss": 0.1543,
+      "step": 7013
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.360640301318268e-05,
+      "loss": 0.1522,
+      "step": 7014
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.360404896421846e-05,
+      "loss": 0.1666,
+      "step": 7015
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.360169491525424e-05,
+      "loss": 0.1389,
+      "step": 7016
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3599340866290016e-05,
+      "loss": 0.0698,
+      "step": 7017
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3596986817325806e-05,
+      "loss": 0.1467,
+      "step": 7018
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3594632768361584e-05,
+      "loss": 0.2533,
+      "step": 7019
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.359227871939736e-05,
+      "loss": 0.0572,
+      "step": 7020
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.358992467043315e-05,
+      "loss": 0.1333,
+      "step": 7021
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.358757062146893e-05,
+      "loss": 0.1743,
+      "step": 7022
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3585216572504706e-05,
+      "loss": 0.2179,
+      "step": 7023
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.358286252354049e-05,
+      "loss": 0.2621,
+      "step": 7024
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3580508474576274e-05,
+      "loss": 0.0942,
+      "step": 7025
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.357815442561205e-05,
+      "loss": 0.1569,
+      "step": 7026
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3575800376647835e-05,
+      "loss": 0.1803,
+      "step": 7027
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.357344632768362e-05,
+      "loss": 0.0304,
+      "step": 7028
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.35710922787194e-05,
+      "loss": 0.1158,
+      "step": 7029
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.356873822975518e-05,
+      "loss": 0.1778,
+      "step": 7030
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3566384180790964e-05,
+      "loss": 0.1996,
+      "step": 7031
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.356403013182675e-05,
+      "loss": 0.2965,
+      "step": 7032
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3561676082862525e-05,
+      "loss": 0.0394,
+      "step": 7033
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.35593220338983e-05,
+      "loss": 0.2156,
+      "step": 7034
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.355696798493409e-05,
+      "loss": 0.1708,
+      "step": 7035
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.355461393596987e-05,
+      "loss": 0.2009,
+      "step": 7036
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.355225988700565e-05,
+      "loss": 0.1221,
+      "step": 7037
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.354990583804144e-05,
+      "loss": 0.2071,
+      "step": 7038
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3547551789077215e-05,
+      "loss": 0.1142,
+      "step": 7039
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.354519774011299e-05,
+      "loss": 0.0766,
+      "step": 7040
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3542843691148776e-05,
+      "loss": 0.1985,
+      "step": 7041
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.354048964218456e-05,
+      "loss": 0.115,
+      "step": 7042
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.353813559322034e-05,
+      "loss": 0.2055,
+      "step": 7043
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.353578154425612e-05,
+      "loss": 0.0599,
+      "step": 7044
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3533427495291905e-05,
+      "loss": 0.2122,
+      "step": 7045
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.353107344632769e-05,
+      "loss": 0.0941,
+      "step": 7046
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3528719397363466e-05,
+      "loss": 0.1161,
+      "step": 7047
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.352636534839925e-05,
+      "loss": 0.1123,
+      "step": 7048
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3524011299435034e-05,
+      "loss": 0.0341,
+      "step": 7049
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.352165725047081e-05,
+      "loss": 0.1817,
+      "step": 7050
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.351930320150659e-05,
+      "loss": 0.2142,
+      "step": 7051
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.351694915254238e-05,
+      "loss": 0.1083,
+      "step": 7052
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3514595103578156e-05,
+      "loss": 0.1297,
+      "step": 7053
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.351224105461393e-05,
+      "loss": 0.1403,
+      "step": 7054
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3509887005649724e-05,
+      "loss": 0.0559,
+      "step": 7055
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.35075329566855e-05,
+      "loss": 0.2052,
+      "step": 7056
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.350517890772128e-05,
+      "loss": 0.0504,
+      "step": 7057
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.350282485875706e-05,
+      "loss": 0.09,
+      "step": 7058
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3500470809792846e-05,
+      "loss": 0.1126,
+      "step": 7059
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.349811676082863e-05,
+      "loss": 0.0714,
+      "step": 7060
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.349576271186441e-05,
+      "loss": 0.2109,
+      "step": 7061
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.349340866290019e-05,
+      "loss": 0.1039,
+      "step": 7062
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3491054613935975e-05,
+      "loss": 0.2297,
+      "step": 7063
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.348870056497175e-05,
+      "loss": 0.0849,
+      "step": 7064
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3486346516007536e-05,
+      "loss": 0.0916,
+      "step": 7065
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.348399246704332e-05,
+      "loss": 0.1456,
+      "step": 7066
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.34816384180791e-05,
+      "loss": 0.1095,
+      "step": 7067
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3479284369114874e-05,
+      "loss": 0.1566,
+      "step": 7068
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3476930320150665e-05,
+      "loss": 0.2859,
+      "step": 7069
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.347457627118644e-05,
+      "loss": 0.0943,
+      "step": 7070
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.347222222222222e-05,
+      "loss": 0.0927,
+      "step": 7071
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.346986817325801e-05,
+      "loss": 0.0711,
+      "step": 7072
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.346751412429379e-05,
+      "loss": 0.1382,
+      "step": 7073
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3465160075329564e-05,
+      "loss": 0.283,
+      "step": 7074
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.346280602636535e-05,
+      "loss": 0.0597,
+      "step": 7075
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.346045197740113e-05,
+      "loss": 0.0786,
+      "step": 7076
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3458097928436916e-05,
+      "loss": 0.1449,
+      "step": 7077
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.345574387947269e-05,
+      "loss": 0.2144,
+      "step": 7078
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.345338983050848e-05,
+      "loss": 0.1638,
+      "step": 7079
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.345103578154426e-05,
+      "loss": 0.0824,
+      "step": 7080
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.344868173258004e-05,
+      "loss": 0.0449,
+      "step": 7081
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.344632768361582e-05,
+      "loss": 0.1275,
+      "step": 7082
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3443973634651606e-05,
+      "loss": 0.1114,
+      "step": 7083
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.344161958568738e-05,
+      "loss": 0.095,
+      "step": 7084
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.343926553672316e-05,
+      "loss": 0.1027,
+      "step": 7085
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.343691148775895e-05,
+      "loss": 0.2238,
+      "step": 7086
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.343455743879473e-05,
+      "loss": 0.1654,
+      "step": 7087
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3432203389830505e-05,
+      "loss": 0.1753,
+      "step": 7088
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3429849340866296e-05,
+      "loss": 0.2059,
+      "step": 7089
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.342749529190207e-05,
+      "loss": 0.105,
+      "step": 7090
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.342514124293786e-05,
+      "loss": 0.1613,
+      "step": 7091
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3422787193973634e-05,
+      "loss": 0.0492,
+      "step": 7092
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.342043314500942e-05,
+      "loss": 0.1412,
+      "step": 7093
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.34180790960452e-05,
+      "loss": 0.1699,
+      "step": 7094
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.341572504708098e-05,
+      "loss": 0.0621,
+      "step": 7095
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.341337099811676e-05,
+      "loss": 0.1009,
+      "step": 7096
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.341101694915255e-05,
+      "loss": 0.0903,
+      "step": 7097
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3408662900188324e-05,
+      "loss": 0.2346,
+      "step": 7098
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.340630885122411e-05,
+      "loss": 0.1856,
+      "step": 7099
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.340395480225989e-05,
+      "loss": 0.0913,
+      "step": 7100
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.340160075329567e-05,
+      "loss": 0.1235,
+      "step": 7101
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.339924670433145e-05,
+      "loss": 0.128,
+      "step": 7102
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.339689265536724e-05,
+      "loss": 0.2845,
+      "step": 7103
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3394538606403014e-05,
+      "loss": 0.1782,
+      "step": 7104
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.339218455743879e-05,
+      "loss": 0.0588,
+      "step": 7105
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.338983050847458e-05,
+      "loss": 0.2617,
+      "step": 7106
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.338747645951036e-05,
+      "loss": 0.0834,
+      "step": 7107
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.338512241054614e-05,
+      "loss": 0.1772,
+      "step": 7108
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.338276836158192e-05,
+      "loss": 0.1261,
+      "step": 7109
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3380414312617704e-05,
+      "loss": 0.1647,
+      "step": 7110
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.337806026365349e-05,
+      "loss": 0.167,
+      "step": 7111
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3375706214689265e-05,
+      "loss": 0.1426,
+      "step": 7112
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.337335216572505e-05,
+      "loss": 0.1202,
+      "step": 7113
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.337099811676083e-05,
+      "loss": 0.0913,
+      "step": 7114
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.336864406779661e-05,
+      "loss": 0.1162,
+      "step": 7115
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3366290018832394e-05,
+      "loss": 0.1144,
+      "step": 7116
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.336393596986818e-05,
+      "loss": 0.1198,
+      "step": 7117
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3361581920903955e-05,
+      "loss": 0.2006,
+      "step": 7118
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.335922787193974e-05,
+      "loss": 0.1013,
+      "step": 7119
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.335687382297552e-05,
+      "loss": 0.1588,
+      "step": 7120
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.33545197740113e-05,
+      "loss": 0.1868,
+      "step": 7121
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.335216572504708e-05,
+      "loss": 0.1076,
+      "step": 7122
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.334981167608287e-05,
+      "loss": 0.1153,
+      "step": 7123
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3347457627118645e-05,
+      "loss": 0.0669,
+      "step": 7124
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.334510357815443e-05,
+      "loss": 0.104,
+      "step": 7125
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3342749529190206e-05,
+      "loss": 0.1438,
+      "step": 7126
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.334039548022599e-05,
+      "loss": 0.291,
+      "step": 7127
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3338041431261774e-05,
+      "loss": 0.0654,
+      "step": 7128
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.333568738229755e-05,
+      "loss": 0.2011,
+      "step": 7129
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.1413,
+      "step": 7130
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.333097928436912e-05,
+      "loss": 0.3233,
+      "step": 7131
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3328625235404896e-05,
+      "loss": 0.0719,
+      "step": 7132
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.332627118644068e-05,
+      "loss": 0.1503,
+      "step": 7133
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3323917137476464e-05,
+      "loss": 0.2028,
+      "step": 7134
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.332156308851224e-05,
+      "loss": 0.2008,
+      "step": 7135
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3319209039548025e-05,
+      "loss": 0.1385,
+      "step": 7136
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.331685499058381e-05,
+      "loss": 0.0815,
+      "step": 7137
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3314500941619586e-05,
+      "loss": 0.0739,
+      "step": 7138
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.331214689265537e-05,
+      "loss": 0.2077,
+      "step": 7139
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3309792843691154e-05,
+      "loss": 0.1606,
+      "step": 7140
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.330743879472693e-05,
+      "loss": 0.1414,
+      "step": 7141
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3305084745762715e-05,
+      "loss": 0.1186,
+      "step": 7142
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.330273069679849e-05,
+      "loss": 0.1478,
+      "step": 7143
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3300376647834276e-05,
+      "loss": 0.1012,
+      "step": 7144
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.329802259887006e-05,
+      "loss": 0.1593,
+      "step": 7145
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.329566854990584e-05,
+      "loss": 0.1238,
+      "step": 7146
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.329331450094162e-05,
+      "loss": 0.1618,
+      "step": 7147
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3290960451977405e-05,
+      "loss": 0.1045,
+      "step": 7148
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.328860640301318e-05,
+      "loss": 0.1472,
+      "step": 7149
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3286252354048966e-05,
+      "loss": 0.0665,
+      "step": 7150
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.328389830508475e-05,
+      "loss": 0.2481,
+      "step": 7151
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.328154425612053e-05,
+      "loss": 0.0734,
+      "step": 7152
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.327919020715631e-05,
+      "loss": 0.1428,
+      "step": 7153
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3276836158192095e-05,
+      "loss": 0.0554,
+      "step": 7154
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.327448210922787e-05,
+      "loss": 0.0434,
+      "step": 7155
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3272128060263656e-05,
+      "loss": 0.1128,
+      "step": 7156
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.326977401129944e-05,
+      "loss": 0.1939,
+      "step": 7157
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.326741996233522e-05,
+      "loss": 0.0585,
+      "step": 7158
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3265065913371e-05,
+      "loss": 0.1338,
+      "step": 7159
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.326271186440678e-05,
+      "loss": 0.2939,
+      "step": 7160
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.326035781544256e-05,
+      "loss": 0.1783,
+      "step": 7161
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3258003766478346e-05,
+      "loss": 0.1667,
+      "step": 7162
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.325564971751412e-05,
+      "loss": 0.2377,
+      "step": 7163
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.325329566854991e-05,
+      "loss": 0.1525,
+      "step": 7164
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.325094161958569e-05,
+      "loss": 0.1441,
+      "step": 7165
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.324858757062147e-05,
+      "loss": 0.0893,
+      "step": 7166
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.324623352165725e-05,
+      "loss": 0.1898,
+      "step": 7167
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3243879472693036e-05,
+      "loss": 0.158,
+      "step": 7168
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.324152542372881e-05,
+      "loss": 0.1283,
+      "step": 7169
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.32391713747646e-05,
+      "loss": 0.1326,
+      "step": 7170
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.323681732580038e-05,
+      "loss": 0.1316,
+      "step": 7171
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.323446327683616e-05,
+      "loss": 0.0966,
+      "step": 7172
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.323210922787194e-05,
+      "loss": 0.0577,
+      "step": 7173
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3229755178907726e-05,
+      "loss": 0.1313,
+      "step": 7174
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.32274011299435e-05,
+      "loss": 0.2295,
+      "step": 7175
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.322504708097929e-05,
+      "loss": 0.0681,
+      "step": 7176
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3222693032015064e-05,
+      "loss": 0.1717,
+      "step": 7177
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.322033898305085e-05,
+      "loss": 0.2193,
+      "step": 7178
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.321798493408663e-05,
+      "loss": 0.1836,
+      "step": 7179
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.321563088512241e-05,
+      "loss": 0.0859,
+      "step": 7180
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.321327683615819e-05,
+      "loss": 0.1494,
+      "step": 7181
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.321092278719398e-05,
+      "loss": 0.1309,
+      "step": 7182
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3208568738229754e-05,
+      "loss": 0.2258,
+      "step": 7183
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.320621468926554e-05,
+      "loss": 0.0989,
+      "step": 7184
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.320386064030132e-05,
+      "loss": 0.1515,
+      "step": 7185
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.32015065913371e-05,
+      "loss": 0.1671,
+      "step": 7186
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.319915254237288e-05,
+      "loss": 0.1371,
+      "step": 7187
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.319679849340867e-05,
+      "loss": 0.1779,
+      "step": 7188
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3194444444444444e-05,
+      "loss": 0.2368,
+      "step": 7189
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.319209039548023e-05,
+      "loss": 0.0759,
+      "step": 7190
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.318973634651601e-05,
+      "loss": 0.1966,
+      "step": 7191
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.318738229755179e-05,
+      "loss": 0.0973,
+      "step": 7192
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.318502824858757e-05,
+      "loss": 0.2243,
+      "step": 7193
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.318267419962336e-05,
+      "loss": 0.0754,
+      "step": 7194
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3180320150659134e-05,
+      "loss": 0.2052,
+      "step": 7195
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.317796610169492e-05,
+      "loss": 0.1272,
+      "step": 7196
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3175612052730695e-05,
+      "loss": 0.0986,
+      "step": 7197
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.317325800376648e-05,
+      "loss": 0.0925,
+      "step": 7198
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.317090395480226e-05,
+      "loss": 0.0844,
+      "step": 7199
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.316854990583804e-05,
+      "loss": 0.2035,
+      "step": 7200
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3166195856873824e-05,
+      "loss": 0.1347,
+      "step": 7201
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.316384180790961e-05,
+      "loss": 0.1355,
+      "step": 7202
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3161487758945385e-05,
+      "loss": 0.1429,
+      "step": 7203
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.315913370998117e-05,
+      "loss": 0.0442,
+      "step": 7204
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.315677966101695e-05,
+      "loss": 0.1481,
+      "step": 7205
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.315442561205273e-05,
+      "loss": 0.1231,
+      "step": 7206
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3152071563088514e-05,
+      "loss": 0.0869,
+      "step": 7207
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.31497175141243e-05,
+      "loss": 0.1577,
+      "step": 7208
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3147363465160075e-05,
+      "loss": 0.2893,
+      "step": 7209
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.314500941619586e-05,
+      "loss": 0.083,
+      "step": 7210
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.314265536723164e-05,
+      "loss": 0.0554,
+      "step": 7211
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.314030131826742e-05,
+      "loss": 0.1127,
+      "step": 7212
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3137947269303204e-05,
+      "loss": 0.1684,
+      "step": 7213
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.313559322033898e-05,
+      "loss": 0.2099,
+      "step": 7214
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3133239171374765e-05,
+      "loss": 0.2522,
+      "step": 7215
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.313088512241055e-05,
+      "loss": 0.1218,
+      "step": 7216
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3128531073446326e-05,
+      "loss": 0.1576,
+      "step": 7217
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.312617702448211e-05,
+      "loss": 0.1288,
+      "step": 7218
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3123822975517894e-05,
+      "loss": 0.174,
+      "step": 7219
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.312146892655367e-05,
+      "loss": 0.1242,
+      "step": 7220
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3119114877589455e-05,
+      "loss": 0.132,
+      "step": 7221
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.311676082862524e-05,
+      "loss": 0.142,
+      "step": 7222
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3114406779661016e-05,
+      "loss": 0.1156,
+      "step": 7223
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.31120527306968e-05,
+      "loss": 0.1648,
+      "step": 7224
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3109698681732584e-05,
+      "loss": 0.0891,
+      "step": 7225
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.310734463276836e-05,
+      "loss": 0.1514,
+      "step": 7226
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3104990583804145e-05,
+      "loss": 0.14,
+      "step": 7227
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.310263653483993e-05,
+      "loss": 0.1554,
+      "step": 7228
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3100282485875706e-05,
+      "loss": 0.1361,
+      "step": 7229
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.309792843691149e-05,
+      "loss": 0.1889,
+      "step": 7230
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.309557438794727e-05,
+      "loss": 0.275,
+      "step": 7231
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.309322033898305e-05,
+      "loss": 0.126,
+      "step": 7232
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3090866290018835e-05,
+      "loss": 0.1204,
+      "step": 7233
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.308851224105461e-05,
+      "loss": 0.1243,
+      "step": 7234
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3086158192090396e-05,
+      "loss": 0.2551,
+      "step": 7235
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.308380414312618e-05,
+      "loss": 0.0791,
+      "step": 7236
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.308145009416196e-05,
+      "loss": 0.1095,
+      "step": 7237
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.307909604519774e-05,
+      "loss": 0.1271,
+      "step": 7238
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3076741996233525e-05,
+      "loss": 0.1463,
+      "step": 7239
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.30743879472693e-05,
+      "loss": 0.0503,
+      "step": 7240
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3072033898305086e-05,
+      "loss": 0.1949,
+      "step": 7241
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.306967984934087e-05,
+      "loss": 0.2743,
+      "step": 7242
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.306732580037665e-05,
+      "loss": 0.1728,
+      "step": 7243
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.306497175141243e-05,
+      "loss": 0.1155,
+      "step": 7244
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3062617702448215e-05,
+      "loss": 0.1967,
+      "step": 7245
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.306026365348399e-05,
+      "loss": 0.1094,
+      "step": 7246
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3057909604519776e-05,
+      "loss": 0.137,
+      "step": 7247
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3055555555555553e-05,
+      "loss": 0.1222,
+      "step": 7248
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3053201506591344e-05,
+      "loss": 0.1802,
+      "step": 7249
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.305084745762712e-05,
+      "loss": 0.1011,
+      "step": 7250
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.30484934086629e-05,
+      "loss": 0.1424,
+      "step": 7251
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.304613935969868e-05,
+      "loss": 0.2149,
+      "step": 7252
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.3043785310734466e-05,
+      "loss": 0.0743,
+      "step": 7253
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.3041431261770243e-05,
+      "loss": 0.0572,
+      "step": 7254
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.303907721280603e-05,
+      "loss": 0.2291,
+      "step": 7255
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.303672316384181e-05,
+      "loss": 0.205,
+      "step": 7256
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.303436911487759e-05,
+      "loss": 0.1207,
+      "step": 7257
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.303201506591337e-05,
+      "loss": 0.1892,
+      "step": 7258
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.3029661016949156e-05,
+      "loss": 0.0623,
+      "step": 7259
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.3027306967984933e-05,
+      "loss": 0.0244,
+      "step": 7260
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.302495291902072e-05,
+      "loss": 0.1274,
+      "step": 7261
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.30225988700565e-05,
+      "loss": 0.2275,
+      "step": 7262
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.302024482109228e-05,
+      "loss": 0.2815,
+      "step": 7263
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.301789077212806e-05,
+      "loss": 0.1316,
+      "step": 7264
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.301553672316384e-05,
+      "loss": 0.1556,
+      "step": 7265
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.301318267419963e-05,
+      "loss": 0.1553,
+      "step": 7266
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.301082862523541e-05,
+      "loss": 0.1483,
+      "step": 7267
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.3008474576271184e-05,
+      "loss": 0.1357,
+      "step": 7268
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.3006120527306975e-05,
+      "loss": 0.1255,
+      "step": 7269
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.300376647834275e-05,
+      "loss": 0.0555,
+      "step": 7270
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.300141242937853e-05,
+      "loss": 0.102,
+      "step": 7271
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2999058380414313e-05,
+      "loss": 0.0603,
+      "step": 7272
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.29967043314501e-05,
+      "loss": 0.1623,
+      "step": 7273
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2994350282485874e-05,
+      "loss": 0.2256,
+      "step": 7274
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.299199623352166e-05,
+      "loss": 0.1311,
+      "step": 7275
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.298964218455744e-05,
+      "loss": 0.14,
+      "step": 7276
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.298728813559322e-05,
+      "loss": 0.2119,
+      "step": 7277
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2984934086629e-05,
+      "loss": 0.1016,
+      "step": 7278
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.298258003766479e-05,
+      "loss": 0.1409,
+      "step": 7279
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2980225988700564e-05,
+      "loss": 0.0964,
+      "step": 7280
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.297787193973635e-05,
+      "loss": 0.25,
+      "step": 7281
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2975517890772126e-05,
+      "loss": 0.3132,
+      "step": 7282
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2973163841807916e-05,
+      "loss": 0.1695,
+      "step": 7283
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.297080979284369e-05,
+      "loss": 0.0668,
+      "step": 7284
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.296845574387947e-05,
+      "loss": 0.0894,
+      "step": 7285
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.296610169491526e-05,
+      "loss": 0.1101,
+      "step": 7286
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.296374764595104e-05,
+      "loss": 0.1523,
+      "step": 7287
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2961393596986816e-05,
+      "loss": 0.0421,
+      "step": 7288
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.29590395480226e-05,
+      "loss": 0.1737,
+      "step": 7289
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.295668549905838e-05,
+      "loss": 0.11,
+      "step": 7290
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.295433145009416e-05,
+      "loss": 0.1735,
+      "step": 7291
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2951977401129944e-05,
+      "loss": 0.0647,
+      "step": 7292
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.294962335216573e-05,
+      "loss": 0.0687,
+      "step": 7293
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2947269303201506e-05,
+      "loss": 0.0987,
+      "step": 7294
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.294491525423729e-05,
+      "loss": 0.1299,
+      "step": 7295
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.294256120527307e-05,
+      "loss": 0.152,
+      "step": 7296
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.294020715630886e-05,
+      "loss": 0.1125,
+      "step": 7297
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2937853107344634e-05,
+      "loss": 0.0556,
+      "step": 7298
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.293549905838041e-05,
+      "loss": 0.0374,
+      "step": 7299
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.29331450094162e-05,
+      "loss": 0.1019,
+      "step": 7300
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.293079096045198e-05,
+      "loss": 0.0959,
+      "step": 7301
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2928436911487757e-05,
+      "loss": 0.1494,
+      "step": 7302
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.292608286252355e-05,
+      "loss": 0.1821,
+      "step": 7303
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2923728813559324e-05,
+      "loss": 0.1121,
+      "step": 7304
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.29213747645951e-05,
+      "loss": 0.1027,
+      "step": 7305
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2919020715630886e-05,
+      "loss": 0.2217,
+      "step": 7306
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.291666666666667e-05,
+      "loss": 0.0899,
+      "step": 7307
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2914312617702447e-05,
+      "loss": 0.1352,
+      "step": 7308
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.291195856873823e-05,
+      "loss": 0.0556,
+      "step": 7309
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2909604519774014e-05,
+      "loss": 0.0983,
+      "step": 7310
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.290725047080979e-05,
+      "loss": 0.2045,
+      "step": 7311
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2904896421845575e-05,
+      "loss": 0.1659,
+      "step": 7312
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.290254237288136e-05,
+      "loss": 0.1632,
+      "step": 7313
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.290018832391714e-05,
+      "loss": 0.123,
+      "step": 7314
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.289783427495292e-05,
+      "loss": 0.2036,
+      "step": 7315
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.28954802259887e-05,
+      "loss": 0.2535,
+      "step": 7316
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.289312617702449e-05,
+      "loss": 0.1401,
+      "step": 7317
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2890772128060265e-05,
+      "loss": 0.2219,
+      "step": 7318
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.288841807909604e-05,
+      "loss": 0.091,
+      "step": 7319
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.288606403013183e-05,
+      "loss": 0.1202,
+      "step": 7320
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.288370998116761e-05,
+      "loss": 0.159,
+      "step": 7321
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.288135593220339e-05,
+      "loss": 0.1535,
+      "step": 7322
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.287900188323917e-05,
+      "loss": 0.0689,
+      "step": 7323
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2876647834274955e-05,
+      "loss": 0.1683,
+      "step": 7324
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.287429378531073e-05,
+      "loss": 0.2353,
+      "step": 7325
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2871939736346517e-05,
+      "loss": 0.2269,
+      "step": 7326
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.28695856873823e-05,
+      "loss": 0.1402,
+      "step": 7327
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2867231638418084e-05,
+      "loss": 0.2991,
+      "step": 7328
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.286487758945386e-05,
+      "loss": 0.0722,
+      "step": 7329
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2862523540489645e-05,
+      "loss": 0.1323,
+      "step": 7330
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.286016949152543e-05,
+      "loss": 0.201,
+      "step": 7331
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2857815442561207e-05,
+      "loss": 0.1305,
+      "step": 7332
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2855461393596984e-05,
+      "loss": 0.1348,
+      "step": 7333
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2853107344632774e-05,
+      "loss": 0.1507,
+      "step": 7334
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.285075329566855e-05,
+      "loss": 0.1657,
+      "step": 7335
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.284839924670433e-05,
+      "loss": 0.0905,
+      "step": 7336
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.284604519774012e-05,
+      "loss": 0.1205,
+      "step": 7337
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2843691148775897e-05,
+      "loss": 0.0914,
+      "step": 7338
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2841337099811674e-05,
+      "loss": 0.0835,
+      "step": 7339
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.283898305084746e-05,
+      "loss": 0.145,
+      "step": 7340
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.283662900188324e-05,
+      "loss": 0.095,
+      "step": 7341
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.283427495291902e-05,
+      "loss": 0.2077,
+      "step": 7342
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.28319209039548e-05,
+      "loss": 0.0839,
+      "step": 7343
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2829566854990587e-05,
+      "loss": 0.0942,
+      "step": 7344
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.282721280602637e-05,
+      "loss": 0.1568,
+      "step": 7345
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.282485875706215e-05,
+      "loss": 0.1433,
+      "step": 7346
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.282250470809793e-05,
+      "loss": 0.147,
+      "step": 7347
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2820150659133715e-05,
+      "loss": 0.0876,
+      "step": 7348
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.281779661016949e-05,
+      "loss": 0.0741,
+      "step": 7349
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.281544256120527e-05,
+      "loss": 0.0875,
+      "step": 7350
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.281308851224106e-05,
+      "loss": 0.0988,
+      "step": 7351
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.281073446327684e-05,
+      "loss": 0.1336,
+      "step": 7352
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2808380414312615e-05,
+      "loss": 0.2019,
+      "step": 7353
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2806026365348405e-05,
+      "loss": 0.1159,
+      "step": 7354
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.280367231638418e-05,
+      "loss": 0.1234,
+      "step": 7355
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.280131826741996e-05,
+      "loss": 0.0903,
+      "step": 7356
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2798964218455744e-05,
+      "loss": 0.202,
+      "step": 7357
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.279661016949153e-05,
+      "loss": 0.1323,
+      "step": 7358
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2794256120527305e-05,
+      "loss": 0.1917,
+      "step": 7359
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.279190207156309e-05,
+      "loss": 0.1835,
+      "step": 7360
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.278954802259887e-05,
+      "loss": 0.1026,
+      "step": 7361
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2787193973634657e-05,
+      "loss": 0.0584,
+      "step": 7362
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2784839924670434e-05,
+      "loss": 0.0841,
+      "step": 7363
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.278248587570622e-05,
+      "loss": 0.1222,
+      "step": 7364
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2780131826742e-05,
+      "loss": 0.1403,
+      "step": 7365
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.277777777777778e-05,
+      "loss": 0.1425,
+      "step": 7366
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2775423728813556e-05,
+      "loss": 0.1634,
+      "step": 7367
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2773069679849347e-05,
+      "loss": 0.1395,
+      "step": 7368
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2770715630885124e-05,
+      "loss": 0.2607,
+      "step": 7369
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.27683615819209e-05,
+      "loss": 0.1736,
+      "step": 7370
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.276600753295669e-05,
+      "loss": 0.0543,
+      "step": 7371
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.276365348399247e-05,
+      "loss": 0.1323,
+      "step": 7372
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2761299435028246e-05,
+      "loss": 0.1969,
+      "step": 7373
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.275894538606403e-05,
+      "loss": 0.1146,
+      "step": 7374
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2756591337099814e-05,
+      "loss": 0.0608,
+      "step": 7375
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.27542372881356e-05,
+      "loss": 0.2116,
+      "step": 7376
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2751883239171375e-05,
+      "loss": 0.1103,
+      "step": 7377
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.274952919020716e-05,
+      "loss": 0.2416,
+      "step": 7378
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.274717514124294e-05,
+      "loss": 0.143,
+      "step": 7379
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.274482109227872e-05,
+      "loss": 0.1486,
+      "step": 7380
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2742467043314504e-05,
+      "loss": 0.1459,
+      "step": 7381
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.274011299435029e-05,
+      "loss": 0.1627,
+      "step": 7382
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2737758945386065e-05,
+      "loss": 0.0583,
+      "step": 7383
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.273540489642184e-05,
+      "loss": 0.1189,
+      "step": 7384
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.273305084745763e-05,
+      "loss": 0.0914,
+      "step": 7385
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.273069679849341e-05,
+      "loss": 0.1145,
+      "step": 7386
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.272834274952919e-05,
+      "loss": 0.0492,
+      "step": 7387
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.272598870056498e-05,
+      "loss": 0.0318,
+      "step": 7388
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2723634651600755e-05,
+      "loss": 0.0999,
+      "step": 7389
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.272128060263653e-05,
+      "loss": 0.1746,
+      "step": 7390
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2718926553672316e-05,
+      "loss": 0.0933,
+      "step": 7391
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.27165725047081e-05,
+      "loss": 0.1552,
+      "step": 7392
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2714218455743884e-05,
+      "loss": 0.0985,
+      "step": 7393
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.271186440677966e-05,
+      "loss": 0.0928,
+      "step": 7394
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2709510357815445e-05,
+      "loss": 0.1592,
+      "step": 7395
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.270715630885123e-05,
+      "loss": 0.1368,
+      "step": 7396
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2704802259887006e-05,
+      "loss": 0.1158,
+      "step": 7397
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.270244821092279e-05,
+      "loss": 0.1711,
+      "step": 7398
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2700094161958574e-05,
+      "loss": 0.1105,
+      "step": 7399
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.269774011299435e-05,
+      "loss": 0.0834,
+      "step": 7400
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.269538606403013e-05,
+      "loss": 0.1248,
+      "step": 7401
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.269303201506592e-05,
+      "loss": 0.2112,
+      "step": 7402
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2690677966101696e-05,
+      "loss": 0.0523,
+      "step": 7403
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.268832391713747e-05,
+      "loss": 0.0786,
+      "step": 7404
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2685969868173264e-05,
+      "loss": 0.1365,
+      "step": 7405
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.268361581920904e-05,
+      "loss": 0.1761,
+      "step": 7406
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2681261770244825e-05,
+      "loss": 0.0308,
+      "step": 7407
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.26789077212806e-05,
+      "loss": 0.1567,
+      "step": 7408
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2676553672316386e-05,
+      "loss": 0.1601,
+      "step": 7409
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.267419962335217e-05,
+      "loss": 0.1616,
+      "step": 7410
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.267184557438795e-05,
+      "loss": 0.07,
+      "step": 7411
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.266949152542373e-05,
+      "loss": 0.0869,
+      "step": 7412
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2667137476459515e-05,
+      "loss": 0.2358,
+      "step": 7413
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.266478342749529e-05,
+      "loss": 0.1124,
+      "step": 7414
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2662429378531076e-05,
+      "loss": 0.046,
+      "step": 7415
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.266007532956686e-05,
+      "loss": 0.1474,
+      "step": 7416
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.265772128060264e-05,
+      "loss": 0.0513,
+      "step": 7417
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2655367231638414e-05,
+      "loss": 0.1713,
+      "step": 7418
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2653013182674205e-05,
+      "loss": 0.1299,
+      "step": 7419
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.265065913370998e-05,
+      "loss": 0.0293,
+      "step": 7420
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.264830508474576e-05,
+      "loss": 0.1124,
+      "step": 7421
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.264595103578155e-05,
+      "loss": 0.0962,
+      "step": 7422
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.264359698681733e-05,
+      "loss": 0.0431,
+      "step": 7423
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.264124293785311e-05,
+      "loss": 0.1717,
+      "step": 7424
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.263888888888889e-05,
+      "loss": 0.1397,
+      "step": 7425
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.263653483992467e-05,
+      "loss": 0.0668,
+      "step": 7426
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2634180790960456e-05,
+      "loss": 0.1049,
+      "step": 7427
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.263182674199623e-05,
+      "loss": 0.2307,
+      "step": 7428
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.262947269303202e-05,
+      "loss": 0.0988,
+      "step": 7429
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.26271186440678e-05,
+      "loss": 0.1358,
+      "step": 7430
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.262476459510358e-05,
+      "loss": 0.1532,
+      "step": 7431
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.262241054613936e-05,
+      "loss": 0.2364,
+      "step": 7432
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2620056497175146e-05,
+      "loss": 0.063,
+      "step": 7433
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.261770244821092e-05,
+      "loss": 0.1507,
+      "step": 7434
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.261534839924671e-05,
+      "loss": 0.1824,
+      "step": 7435
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.261299435028249e-05,
+      "loss": 0.2304,
+      "step": 7436
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.261064030131827e-05,
+      "loss": 0.2792,
+      "step": 7437
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2608286252354045e-05,
+      "loss": 0.09,
+      "step": 7438
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2605932203389836e-05,
+      "loss": 0.1699,
+      "step": 7439
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.260357815442561e-05,
+      "loss": 0.0911,
+      "step": 7440
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.26012241054614e-05,
+      "loss": 0.0346,
+      "step": 7441
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2598870056497174e-05,
+      "loss": 0.0962,
+      "step": 7442
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.259651600753296e-05,
+      "loss": 0.1135,
+      "step": 7443
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.259416195856874e-05,
+      "loss": 0.0739,
+      "step": 7444
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.259180790960452e-05,
+      "loss": 0.0754,
+      "step": 7445
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.25894538606403e-05,
+      "loss": 0.1691,
+      "step": 7446
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.258709981167609e-05,
+      "loss": 0.1457,
+      "step": 7447
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2584745762711864e-05,
+      "loss": 0.0829,
+      "step": 7448
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.258239171374765e-05,
+      "loss": 0.1019,
+      "step": 7449
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.258003766478343e-05,
+      "loss": 0.0941,
+      "step": 7450
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.257768361581921e-05,
+      "loss": 0.2333,
+      "step": 7451
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.257532956685499e-05,
+      "loss": 0.0767,
+      "step": 7452
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.257297551789078e-05,
+      "loss": 0.2472,
+      "step": 7453
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2570621468926554e-05,
+      "loss": 0.0797,
+      "step": 7454
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.256826741996234e-05,
+      "loss": 0.2905,
+      "step": 7455
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.256591337099812e-05,
+      "loss": 0.1141,
+      "step": 7456
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.25635593220339e-05,
+      "loss": 0.094,
+      "step": 7457
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.256120527306968e-05,
+      "loss": 0.2298,
+      "step": 7458
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.255885122410546e-05,
+      "loss": 0.1372,
+      "step": 7459
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2556497175141244e-05,
+      "loss": 0.115,
+      "step": 7460
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.255414312617703e-05,
+      "loss": 0.2194,
+      "step": 7461
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2551789077212805e-05,
+      "loss": 0.0736,
+      "step": 7462
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.254943502824859e-05,
+      "loss": 0.0698,
+      "step": 7463
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.254708097928437e-05,
+      "loss": 0.1315,
+      "step": 7464
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.254472693032015e-05,
+      "loss": 0.1236,
+      "step": 7465
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2542372881355934e-05,
+      "loss": 0.0518,
+      "step": 7466
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.254001883239172e-05,
+      "loss": 0.0886,
+      "step": 7467
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2537664783427495e-05,
+      "loss": 0.1043,
+      "step": 7468
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.253531073446328e-05,
+      "loss": 0.0557,
+      "step": 7469
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.253295668549906e-05,
+      "loss": 0.0998,
+      "step": 7470
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.253060263653484e-05,
+      "loss": 0.1169,
+      "step": 7471
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2528248587570624e-05,
+      "loss": 0.0396,
+      "step": 7472
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.252589453860641e-05,
+      "loss": 0.0654,
+      "step": 7473
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2523540489642185e-05,
+      "loss": 0.1416,
+      "step": 7474
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.252118644067797e-05,
+      "loss": 0.2992,
+      "step": 7475
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2518832391713746e-05,
+      "loss": 0.0879,
+      "step": 7476
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.251647834274953e-05,
+      "loss": 0.2471,
+      "step": 7477
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2514124293785314e-05,
+      "loss": 0.1011,
+      "step": 7478
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.251177024482109e-05,
+      "loss": 0.1602,
+      "step": 7479
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2509416195856875e-05,
+      "loss": 0.2545,
+      "step": 7480
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.250706214689266e-05,
+      "loss": 0.1059,
+      "step": 7481
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2504708097928436e-05,
+      "loss": 0.0956,
+      "step": 7482
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.250235404896422e-05,
+      "loss": 0.1712,
+      "step": 7483
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2500000000000004e-05,
+      "loss": 0.0658,
+      "step": 7484
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.249764595103578e-05,
+      "loss": 0.0988,
+      "step": 7485
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2495291902071565e-05,
+      "loss": 0.1947,
+      "step": 7486
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.249293785310735e-05,
+      "loss": 0.0471,
+      "step": 7487
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2490583804143126e-05,
+      "loss": 0.1741,
+      "step": 7488
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.248822975517891e-05,
+      "loss": 0.1299,
+      "step": 7489
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2485875706214694e-05,
+      "loss": 0.0527,
+      "step": 7490
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.248352165725047e-05,
+      "loss": 0.1396,
+      "step": 7491
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2481167608286255e-05,
+      "loss": 0.0144,
+      "step": 7492
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.247881355932203e-05,
+      "loss": 0.1871,
+      "step": 7493
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2476459510357816e-05,
+      "loss": 0.0978,
+      "step": 7494
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.24741054613936e-05,
+      "loss": 0.2488,
+      "step": 7495
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.247175141242938e-05,
+      "loss": 0.1891,
+      "step": 7496
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.246939736346516e-05,
+      "loss": 0.2354,
+      "step": 7497
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2467043314500945e-05,
+      "loss": 0.2251,
+      "step": 7498
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.246468926553672e-05,
+      "loss": 0.1731,
+      "step": 7499
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2462335216572506e-05,
+      "loss": 0.1342,
+      "step": 7500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.245998116760829e-05,
+      "loss": 0.1969,
+      "step": 7501
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.245762711864407e-05,
+      "loss": 0.1795,
+      "step": 7502
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.245527306967985e-05,
+      "loss": 0.1937,
+      "step": 7503
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2452919020715635e-05,
+      "loss": 0.0677,
+      "step": 7504
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.245056497175141e-05,
+      "loss": 0.2445,
+      "step": 7505
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2448210922787196e-05,
+      "loss": 0.0556,
+      "step": 7506
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.244585687382298e-05,
+      "loss": 0.0827,
+      "step": 7507
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.244350282485876e-05,
+      "loss": 0.09,
+      "step": 7508
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.244114877589454e-05,
+      "loss": 0.1183,
+      "step": 7509
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2438794726930325e-05,
+      "loss": 0.0588,
+      "step": 7510
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.24364406779661e-05,
+      "loss": 0.1156,
+      "step": 7511
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2434086629001886e-05,
+      "loss": 0.0674,
+      "step": 7512
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.243173258003766e-05,
+      "loss": 0.2991,
+      "step": 7513
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.242937853107345e-05,
+      "loss": 0.0412,
+      "step": 7514
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.242702448210923e-05,
+      "loss": 0.1552,
+      "step": 7515
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.242467043314501e-05,
+      "loss": 0.0963,
+      "step": 7516
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.242231638418079e-05,
+      "loss": 0.093,
+      "step": 7517
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2419962335216576e-05,
+      "loss": 0.0721,
+      "step": 7518
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.241760828625235e-05,
+      "loss": 0.3185,
+      "step": 7519
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.241525423728814e-05,
+      "loss": 0.1304,
+      "step": 7520
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.241290018832392e-05,
+      "loss": 0.1645,
+      "step": 7521
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.24105461393597e-05,
+      "loss": 0.0402,
+      "step": 7522
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.240819209039548e-05,
+      "loss": 0.0979,
+      "step": 7523
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2405838041431266e-05,
+      "loss": 0.1621,
+      "step": 7524
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.240348399246704e-05,
+      "loss": 0.1741,
+      "step": 7525
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.240112994350283e-05,
+      "loss": 0.2165,
+      "step": 7526
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.239877589453861e-05,
+      "loss": 0.1971,
+      "step": 7527
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.239642184557439e-05,
+      "loss": 0.2444,
+      "step": 7528
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.239406779661017e-05,
+      "loss": 0.1299,
+      "step": 7529
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.239171374764595e-05,
+      "loss": 0.0868,
+      "step": 7530
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.238935969868173e-05,
+      "loss": 0.0763,
+      "step": 7531
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.238700564971752e-05,
+      "loss": 0.0753,
+      "step": 7532
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2384651600753294e-05,
+      "loss": 0.0712,
+      "step": 7533
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.238229755178908e-05,
+      "loss": 0.1478,
+      "step": 7534
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.237994350282486e-05,
+      "loss": 0.0779,
+      "step": 7535
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.237758945386064e-05,
+      "loss": 0.0705,
+      "step": 7536
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.237523540489642e-05,
+      "loss": 0.1836,
+      "step": 7537
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.237288135593221e-05,
+      "loss": 0.1354,
+      "step": 7538
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2370527306967984e-05,
+      "loss": 0.0942,
+      "step": 7539
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.236817325800377e-05,
+      "loss": 0.051,
+      "step": 7540
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.236581920903955e-05,
+      "loss": 0.0465,
+      "step": 7541
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.236346516007533e-05,
+      "loss": 0.2117,
+      "step": 7542
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.236111111111111e-05,
+      "loss": 0.1299,
+      "step": 7543
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.23587570621469e-05,
+      "loss": 0.1866,
+      "step": 7544
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2356403013182674e-05,
+      "loss": 0.2516,
+      "step": 7545
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.235404896421846e-05,
+      "loss": 0.0806,
+      "step": 7546
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2351694915254235e-05,
+      "loss": 0.133,
+      "step": 7547
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.234934086629002e-05,
+      "loss": 0.0887,
+      "step": 7548
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.23469868173258e-05,
+      "loss": 0.0687,
+      "step": 7549
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.234463276836158e-05,
+      "loss": 0.0661,
+      "step": 7550
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2342278719397364e-05,
+      "loss": 0.0765,
+      "step": 7551
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.233992467043315e-05,
+      "loss": 0.1196,
+      "step": 7552
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2337570621468925e-05,
+      "loss": 0.2505,
+      "step": 7553
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.233521657250471e-05,
+      "loss": 0.1526,
+      "step": 7554
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.233286252354049e-05,
+      "loss": 0.173,
+      "step": 7555
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.233050847457627e-05,
+      "loss": 0.1276,
+      "step": 7556
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2328154425612054e-05,
+      "loss": 0.1471,
+      "step": 7557
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.232580037664784e-05,
+      "loss": 0.1438,
+      "step": 7558
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2323446327683615e-05,
+      "loss": 0.1157,
+      "step": 7559
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.23210922787194e-05,
+      "loss": 0.0791,
+      "step": 7560
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.231873822975518e-05,
+      "loss": 0.1248,
+      "step": 7561
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.231638418079096e-05,
+      "loss": 0.1771,
+      "step": 7562
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2314030131826744e-05,
+      "loss": 0.1423,
+      "step": 7563
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.231167608286252e-05,
+      "loss": 0.091,
+      "step": 7564
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.230932203389831e-05,
+      "loss": 0.1336,
+      "step": 7565
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.230696798493409e-05,
+      "loss": 0.2367,
+      "step": 7566
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2304613935969866e-05,
+      "loss": 0.0562,
+      "step": 7567
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.230225988700565e-05,
+      "loss": 0.1444,
+      "step": 7568
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2299905838041434e-05,
+      "loss": 0.1087,
+      "step": 7569
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.229755178907721e-05,
+      "loss": 0.1025,
+      "step": 7570
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2295197740112995e-05,
+      "loss": 0.1422,
+      "step": 7571
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.229284369114878e-05,
+      "loss": 0.1457,
+      "step": 7572
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2290489642184556e-05,
+      "loss": 0.233,
+      "step": 7573
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.228813559322034e-05,
+      "loss": 0.2056,
+      "step": 7574
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2285781544256124e-05,
+      "loss": 0.1578,
+      "step": 7575
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.22834274952919e-05,
+      "loss": 0.0897,
+      "step": 7576
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2281073446327685e-05,
+      "loss": 0.1162,
+      "step": 7577
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.227871939736347e-05,
+      "loss": 0.1469,
+      "step": 7578
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2276365348399246e-05,
+      "loss": 0.1168,
+      "step": 7579
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.227401129943503e-05,
+      "loss": 0.1349,
+      "step": 7580
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.227165725047081e-05,
+      "loss": 0.2471,
+      "step": 7581
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.22693032015066e-05,
+      "loss": 0.103,
+      "step": 7582
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2266949152542375e-05,
+      "loss": 0.0924,
+      "step": 7583
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.226459510357815e-05,
+      "loss": 0.0723,
+      "step": 7584
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.226224105461394e-05,
+      "loss": 0.1789,
+      "step": 7585
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.225988700564972e-05,
+      "loss": 0.0485,
+      "step": 7586
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.22575329566855e-05,
+      "loss": 0.0747,
+      "step": 7587
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.225517890772128e-05,
+      "loss": 0.0562,
+      "step": 7588
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2252824858757065e-05,
+      "loss": 0.0942,
+      "step": 7589
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.225047080979284e-05,
+      "loss": 0.1041,
+      "step": 7590
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2248116760828626e-05,
+      "loss": 0.0215,
+      "step": 7591
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.224576271186441e-05,
+      "loss": 0.196,
+      "step": 7592
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.224340866290019e-05,
+      "loss": 0.1705,
+      "step": 7593
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.224105461393597e-05,
+      "loss": 0.0868,
+      "step": 7594
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2238700564971755e-05,
+      "loss": 0.2561,
+      "step": 7595
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.223634651600753e-05,
+      "loss": 0.0671,
+      "step": 7596
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2233992467043316e-05,
+      "loss": 0.1773,
+      "step": 7597
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2231638418079093e-05,
+      "loss": 0.3296,
+      "step": 7598
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2229284369114884e-05,
+      "loss": 0.3105,
+      "step": 7599
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.222693032015066e-05,
+      "loss": 0.1401,
+      "step": 7600
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.222457627118644e-05,
+      "loss": 0.19,
+      "step": 7601
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.222222222222223e-05,
+      "loss": 0.1141,
+      "step": 7602
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2219868173258006e-05,
+      "loss": 0.2191,
+      "step": 7603
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2217514124293783e-05,
+      "loss": 0.1262,
+      "step": 7604
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.221516007532957e-05,
+      "loss": 0.103,
+      "step": 7605
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.221280602636535e-05,
+      "loss": 0.1365,
+      "step": 7606
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.221045197740113e-05,
+      "loss": 0.1039,
+      "step": 7607
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.220809792843691e-05,
+      "loss": 0.1996,
+      "step": 7608
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2205743879472696e-05,
+      "loss": 0.0661,
+      "step": 7609
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2203389830508473e-05,
+      "loss": 0.1918,
+      "step": 7610
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.220103578154426e-05,
+      "loss": 0.1165,
+      "step": 7611
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.219868173258004e-05,
+      "loss": 0.1435,
+      "step": 7612
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2196327683615825e-05,
+      "loss": 0.1586,
+      "step": 7613
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.21939736346516e-05,
+      "loss": 0.2235,
+      "step": 7614
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.219161958568738e-05,
+      "loss": 0.1802,
+      "step": 7615
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.218926553672317e-05,
+      "loss": 0.2048,
+      "step": 7616
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.218691148775895e-05,
+      "loss": 0.2625,
+      "step": 7617
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2184557438794725e-05,
+      "loss": 0.0906,
+      "step": 7618
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2182203389830515e-05,
+      "loss": 0.0916,
+      "step": 7619
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.217984934086629e-05,
+      "loss": 0.2081,
+      "step": 7620
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.217749529190207e-05,
+      "loss": 0.079,
+      "step": 7621
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2175141242937853e-05,
+      "loss": 0.067,
+      "step": 7622
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.217278719397364e-05,
+      "loss": 0.1691,
+      "step": 7623
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2170433145009415e-05,
+      "loss": 0.1612,
+      "step": 7624
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.21680790960452e-05,
+      "loss": 0.1147,
+      "step": 7625
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.216572504708098e-05,
+      "loss": 0.1381,
+      "step": 7626
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.216337099811676e-05,
+      "loss": 0.0765,
+      "step": 7627
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2161016949152543e-05,
+      "loss": 0.054,
+      "step": 7628
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.215866290018833e-05,
+      "loss": 0.1939,
+      "step": 7629
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.215630885122411e-05,
+      "loss": 0.1618,
+      "step": 7630
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.215395480225989e-05,
+      "loss": 0.1669,
+      "step": 7631
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2151600753295666e-05,
+      "loss": 0.1786,
+      "step": 7632
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2149246704331456e-05,
+      "loss": 0.1265,
+      "step": 7633
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2146892655367233e-05,
+      "loss": 0.2427,
+      "step": 7634
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.214453860640301e-05,
+      "loss": 0.1971,
+      "step": 7635
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.21421845574388e-05,
+      "loss": 0.1251,
+      "step": 7636
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.213983050847458e-05,
+      "loss": 0.2004,
+      "step": 7637
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2137476459510356e-05,
+      "loss": 0.1652,
+      "step": 7638
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.213512241054614e-05,
+      "loss": 0.0641,
+      "step": 7639
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2132768361581923e-05,
+      "loss": 0.1115,
+      "step": 7640
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.21304143126177e-05,
+      "loss": 0.1835,
+      "step": 7641
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2128060263653485e-05,
+      "loss": 0.1645,
+      "step": 7642
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.212570621468927e-05,
+      "loss": 0.1147,
+      "step": 7643
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.212335216572505e-05,
+      "loss": 0.0682,
+      "step": 7644
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.212099811676083e-05,
+      "loss": 0.1296,
+      "step": 7645
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2118644067796613e-05,
+      "loss": 0.134,
+      "step": 7646
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.21162900188324e-05,
+      "loss": 0.1473,
+      "step": 7647
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2113935969868175e-05,
+      "loss": 0.1906,
+      "step": 7648
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.211158192090395e-05,
+      "loss": 0.1007,
+      "step": 7649
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.210922787193974e-05,
+      "loss": 0.0613,
+      "step": 7650
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.210687382297552e-05,
+      "loss": 0.0208,
+      "step": 7651
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.21045197740113e-05,
+      "loss": 0.0591,
+      "step": 7652
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.210216572504709e-05,
+      "loss": 0.1597,
+      "step": 7653
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2099811676082865e-05,
+      "loss": 0.1158,
+      "step": 7654
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.209745762711864e-05,
+      "loss": 0.1082,
+      "step": 7655
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2095103578154426e-05,
+      "loss": 0.2632,
+      "step": 7656
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.209274952919021e-05,
+      "loss": 0.1671,
+      "step": 7657
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.209039548022599e-05,
+      "loss": 0.1748,
+      "step": 7658
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.208804143126177e-05,
+      "loss": 0.2584,
+      "step": 7659
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2085687382297555e-05,
+      "loss": 0.1567,
+      "step": 7660
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.208333333333334e-05,
+      "loss": 0.0529,
+      "step": 7661
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2080979284369116e-05,
+      "loss": 0.0643,
+      "step": 7662
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.20786252354049e-05,
+      "loss": 0.077,
+      "step": 7663
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2076271186440683e-05,
+      "loss": 0.3466,
+      "step": 7664
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.207391713747646e-05,
+      "loss": 0.2688,
+      "step": 7665
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.207156308851224e-05,
+      "loss": 0.0843,
+      "step": 7666
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.206920903954803e-05,
+      "loss": 0.0558,
+      "step": 7667
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2066854990583806e-05,
+      "loss": 0.1171,
+      "step": 7668
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.206450094161958e-05,
+      "loss": 0.2168,
+      "step": 7669
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2062146892655373e-05,
+      "loss": 0.0866,
+      "step": 7670
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.205979284369115e-05,
+      "loss": 0.089,
+      "step": 7671
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.205743879472693e-05,
+      "loss": 0.2144,
+      "step": 7672
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.205508474576271e-05,
+      "loss": 0.0936,
+      "step": 7673
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2052730696798496e-05,
+      "loss": 0.2765,
+      "step": 7674
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.205037664783427e-05,
+      "loss": 0.1649,
+      "step": 7675
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.204802259887006e-05,
+      "loss": 0.0991,
+      "step": 7676
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.204566854990584e-05,
+      "loss": 0.1607,
+      "step": 7677
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.2043314500941624e-05,
+      "loss": 0.1484,
+      "step": 7678
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.20409604519774e-05,
+      "loss": 0.1881,
+      "step": 7679
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.2038606403013186e-05,
+      "loss": 0.1055,
+      "step": 7680
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.203625235404897e-05,
+      "loss": 0.1254,
+      "step": 7681
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.203389830508475e-05,
+      "loss": 0.1859,
+      "step": 7682
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.2031544256120524e-05,
+      "loss": 0.1109,
+      "step": 7683
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.2029190207156314e-05,
+      "loss": 0.1291,
+      "step": 7684
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.202683615819209e-05,
+      "loss": 0.1971,
+      "step": 7685
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.202448210922787e-05,
+      "loss": 0.1601,
+      "step": 7686
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.202212806026366e-05,
+      "loss": 0.0522,
+      "step": 7687
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.201977401129944e-05,
+      "loss": 0.1599,
+      "step": 7688
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.2017419962335214e-05,
+      "loss": 0.0555,
+      "step": 7689
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.2015065913371e-05,
+      "loss": 0.0849,
+      "step": 7690
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.201271186440678e-05,
+      "loss": 0.1361,
+      "step": 7691
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.2010357815442566e-05,
+      "loss": 0.1511,
+      "step": 7692
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.200800376647834e-05,
+      "loss": 0.079,
+      "step": 7693
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.200564971751413e-05,
+      "loss": 0.0899,
+      "step": 7694
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.200329566854991e-05,
+      "loss": 0.1478,
+      "step": 7695
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.200094161958569e-05,
+      "loss": 0.0663,
+      "step": 7696
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.199858757062147e-05,
+      "loss": 0.1263,
+      "step": 7697
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1996233521657256e-05,
+      "loss": 0.1359,
+      "step": 7698
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.199387947269303e-05,
+      "loss": 0.1674,
+      "step": 7699
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.199152542372881e-05,
+      "loss": 0.1706,
+      "step": 7700
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.19891713747646e-05,
+      "loss": 0.0639,
+      "step": 7701
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.198681732580038e-05,
+      "loss": 0.0401,
+      "step": 7702
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1984463276836155e-05,
+      "loss": 0.1596,
+      "step": 7703
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1982109227871946e-05,
+      "loss": 0.1281,
+      "step": 7704
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.197975517890772e-05,
+      "loss": 0.1984,
+      "step": 7705
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.19774011299435e-05,
+      "loss": 0.2212,
+      "step": 7706
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1975047080979284e-05,
+      "loss": 0.1189,
+      "step": 7707
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.197269303201507e-05,
+      "loss": 0.1547,
+      "step": 7708
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.197033898305085e-05,
+      "loss": 0.1288,
+      "step": 7709
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.196798493408663e-05,
+      "loss": 0.0611,
+      "step": 7710
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.196563088512241e-05,
+      "loss": 0.1088,
+      "step": 7711
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1963276836158197e-05,
+      "loss": 0.1398,
+      "step": 7712
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1960922787193974e-05,
+      "loss": 0.1308,
+      "step": 7713
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.195856873822976e-05,
+      "loss": 0.0976,
+      "step": 7714
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.195621468926554e-05,
+      "loss": 0.0676,
+      "step": 7715
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.195386064030132e-05,
+      "loss": 0.1486,
+      "step": 7716
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1951506591337096e-05,
+      "loss": 0.0567,
+      "step": 7717
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1949152542372887e-05,
+      "loss": 0.0813,
+      "step": 7718
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1946798493408664e-05,
+      "loss": 0.2973,
+      "step": 7719
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.194444444444444e-05,
+      "loss": 0.1183,
+      "step": 7720
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.194209039548023e-05,
+      "loss": 0.0467,
+      "step": 7721
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.193973634651601e-05,
+      "loss": 0.2175,
+      "step": 7722
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.193738229755179e-05,
+      "loss": 0.1665,
+      "step": 7723
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.193502824858757e-05,
+      "loss": 0.096,
+      "step": 7724
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1932674199623354e-05,
+      "loss": 0.1074,
+      "step": 7725
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.193032015065914e-05,
+      "loss": 0.1468,
+      "step": 7726
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1927966101694915e-05,
+      "loss": 0.0406,
+      "step": 7727
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.19256120527307e-05,
+      "loss": 0.1488,
+      "step": 7728
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.192325800376648e-05,
+      "loss": 0.0757,
+      "step": 7729
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.192090395480226e-05,
+      "loss": 0.0842,
+      "step": 7730
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1918549905838044e-05,
+      "loss": 0.2154,
+      "step": 7731
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.191619585687383e-05,
+      "loss": 0.1008,
+      "step": 7732
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1913841807909605e-05,
+      "loss": 0.0894,
+      "step": 7733
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.191148775894538e-05,
+      "loss": 0.2623,
+      "step": 7734
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.190913370998117e-05,
+      "loss": 0.0503,
+      "step": 7735
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.190677966101695e-05,
+      "loss": 0.1041,
+      "step": 7736
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.190442561205273e-05,
+      "loss": 0.2086,
+      "step": 7737
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.190207156308852e-05,
+      "loss": 0.0413,
+      "step": 7738
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1899717514124295e-05,
+      "loss": 0.0537,
+      "step": 7739
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.189736346516008e-05,
+      "loss": 0.1775,
+      "step": 7740
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1895009416195856e-05,
+      "loss": 0.1277,
+      "step": 7741
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.189265536723164e-05,
+      "loss": 0.1599,
+      "step": 7742
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1890301318267424e-05,
+      "loss": 0.1,
+      "step": 7743
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.18879472693032e-05,
+      "loss": 0.1072,
+      "step": 7744
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1885593220338985e-05,
+      "loss": 0.1049,
+      "step": 7745
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.188323917137477e-05,
+      "loss": 0.2572,
+      "step": 7746
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1880885122410546e-05,
+      "loss": 0.0939,
+      "step": 7747
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.187853107344633e-05,
+      "loss": 0.1123,
+      "step": 7748
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1876177024482114e-05,
+      "loss": 0.2157,
+      "step": 7749
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.187382297551789e-05,
+      "loss": 0.2316,
+      "step": 7750
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1871468926553675e-05,
+      "loss": 0.1288,
+      "step": 7751
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.186911487758946e-05,
+      "loss": 0.1094,
+      "step": 7752
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1866760828625236e-05,
+      "loss": 0.1616,
+      "step": 7753
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.186440677966101e-05,
+      "loss": 0.1613,
+      "step": 7754
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1862052730696804e-05,
+      "loss": 0.2338,
+      "step": 7755
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.185969868173258e-05,
+      "loss": 0.1355,
+      "step": 7756
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1857344632768365e-05,
+      "loss": 0.0525,
+      "step": 7757
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.185499058380414e-05,
+      "loss": 0.2135,
+      "step": 7758
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1852636534839926e-05,
+      "loss": 0.0746,
+      "step": 7759
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.185028248587571e-05,
+      "loss": 0.2331,
+      "step": 7760
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.184792843691149e-05,
+      "loss": 0.137,
+      "step": 7761
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.184557438794727e-05,
+      "loss": 0.1814,
+      "step": 7762
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1843220338983055e-05,
+      "loss": 0.0666,
+      "step": 7763
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.184086629001883e-05,
+      "loss": 0.1167,
+      "step": 7764
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1838512241054616e-05,
+      "loss": 0.084,
+      "step": 7765
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.18361581920904e-05,
+      "loss": 0.1472,
+      "step": 7766
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.183380414312618e-05,
+      "loss": 0.1498,
+      "step": 7767
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.183145009416196e-05,
+      "loss": 0.1283,
+      "step": 7768
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1829096045197745e-05,
+      "loss": 0.1898,
+      "step": 7769
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.182674199623352e-05,
+      "loss": 0.1537,
+      "step": 7770
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1824387947269306e-05,
+      "loss": 0.21,
+      "step": 7771
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.182203389830509e-05,
+      "loss": 0.0549,
+      "step": 7772
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.181967984934087e-05,
+      "loss": 0.2405,
+      "step": 7773
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.181732580037665e-05,
+      "loss": 0.1898,
+      "step": 7774
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.181497175141243e-05,
+      "loss": 0.1384,
+      "step": 7775
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.181261770244821e-05,
+      "loss": 0.0679,
+      "step": 7776
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1810263653483996e-05,
+      "loss": 0.1111,
+      "step": 7777
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.180790960451977e-05,
+      "loss": 0.1856,
+      "step": 7778
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.180555555555556e-05,
+      "loss": 0.1681,
+      "step": 7779
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.180320150659134e-05,
+      "loss": 0.1503,
+      "step": 7780
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.180084745762712e-05,
+      "loss": 0.1382,
+      "step": 7781
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.17984934086629e-05,
+      "loss": 0.0699,
+      "step": 7782
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1796139359698686e-05,
+      "loss": 0.1945,
+      "step": 7783
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.179378531073446e-05,
+      "loss": 0.0441,
+      "step": 7784
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.179143126177025e-05,
+      "loss": 0.1221,
+      "step": 7785
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.178907721280603e-05,
+      "loss": 0.1015,
+      "step": 7786
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.178672316384181e-05,
+      "loss": 0.0784,
+      "step": 7787
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.178436911487759e-05,
+      "loss": 0.1719,
+      "step": 7788
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1782015065913376e-05,
+      "loss": 0.0577,
+      "step": 7789
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.177966101694915e-05,
+      "loss": 0.081,
+      "step": 7790
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.177730696798494e-05,
+      "loss": 0.2013,
+      "step": 7791
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1774952919020714e-05,
+      "loss": 0.0914,
+      "step": 7792
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.17725988700565e-05,
+      "loss": 0.1705,
+      "step": 7793
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.177024482109228e-05,
+      "loss": 0.1616,
+      "step": 7794
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.176789077212806e-05,
+      "loss": 0.2354,
+      "step": 7795
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.176553672316384e-05,
+      "loss": 0.0866,
+      "step": 7796
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.176318267419963e-05,
+      "loss": 0.1395,
+      "step": 7797
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1760828625235404e-05,
+      "loss": 0.1063,
+      "step": 7798
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.175847457627119e-05,
+      "loss": 0.0873,
+      "step": 7799
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.175612052730697e-05,
+      "loss": 0.1139,
+      "step": 7800
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.175376647834275e-05,
+      "loss": 0.0603,
+      "step": 7801
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.175141242937853e-05,
+      "loss": 0.1249,
+      "step": 7802
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.174905838041432e-05,
+      "loss": 0.0763,
+      "step": 7803
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1746704331450094e-05,
+      "loss": 0.1718,
+      "step": 7804
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.174435028248588e-05,
+      "loss": 0.0953,
+      "step": 7805
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.174199623352166e-05,
+      "loss": 0.1909,
+      "step": 7806
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.173964218455744e-05,
+      "loss": 0.074,
+      "step": 7807
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.173728813559322e-05,
+      "loss": 0.175,
+      "step": 7808
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1734934086629e-05,
+      "loss": 0.1967,
+      "step": 7809
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1732580037664784e-05,
+      "loss": 0.1123,
+      "step": 7810
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.173022598870057e-05,
+      "loss": 0.139,
+      "step": 7811
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1727871939736345e-05,
+      "loss": 0.1782,
+      "step": 7812
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.172551789077213e-05,
+      "loss": 0.1277,
+      "step": 7813
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.172316384180791e-05,
+      "loss": 0.0721,
+      "step": 7814
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.172080979284369e-05,
+      "loss": 0.0743,
+      "step": 7815
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1718455743879474e-05,
+      "loss": 0.0857,
+      "step": 7816
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.171610169491526e-05,
+      "loss": 0.0644,
+      "step": 7817
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1713747645951035e-05,
+      "loss": 0.1039,
+      "step": 7818
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.171139359698682e-05,
+      "loss": 0.2055,
+      "step": 7819
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.17090395480226e-05,
+      "loss": 0.0732,
+      "step": 7820
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.170668549905838e-05,
+      "loss": 0.1791,
+      "step": 7821
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1704331450094164e-05,
+      "loss": 0.0978,
+      "step": 7822
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.170197740112995e-05,
+      "loss": 0.0528,
+      "step": 7823
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1699623352165725e-05,
+      "loss": 0.1004,
+      "step": 7824
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.169726930320151e-05,
+      "loss": 0.1334,
+      "step": 7825
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.169491525423729e-05,
+      "loss": 0.208,
+      "step": 7826
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.169256120527307e-05,
+      "loss": 0.0634,
+      "step": 7827
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1690207156308854e-05,
+      "loss": 0.0832,
+      "step": 7828
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.168785310734463e-05,
+      "loss": 0.1324,
+      "step": 7829
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1685499058380415e-05,
+      "loss": 0.2711,
+      "step": 7830
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.16831450094162e-05,
+      "loss": 0.0817,
+      "step": 7831
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1680790960451976e-05,
+      "loss": 0.1816,
+      "step": 7832
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.167843691148776e-05,
+      "loss": 0.0856,
+      "step": 7833
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1676082862523544e-05,
+      "loss": 0.041,
+      "step": 7834
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.167372881355932e-05,
+      "loss": 0.1364,
+      "step": 7835
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1671374764595105e-05,
+      "loss": 0.107,
+      "step": 7836
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.166902071563089e-05,
+      "loss": 0.1383,
+      "step": 7837
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1666666666666666e-05,
+      "loss": 0.0993,
+      "step": 7838
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.166431261770245e-05,
+      "loss": 0.1302,
+      "step": 7839
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1661958568738234e-05,
+      "loss": 0.1909,
+      "step": 7840
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.165960451977401e-05,
+      "loss": 0.2161,
+      "step": 7841
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1657250470809795e-05,
+      "loss": 0.0814,
+      "step": 7842
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.165489642184558e-05,
+      "loss": 0.1847,
+      "step": 7843
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1652542372881356e-05,
+      "loss": 0.0759,
+      "step": 7844
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.165018832391714e-05,
+      "loss": 0.0954,
+      "step": 7845
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.164783427495292e-05,
+      "loss": 0.072,
+      "step": 7846
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.16454802259887e-05,
+      "loss": 0.0824,
+      "step": 7847
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1643126177024485e-05,
+      "loss": 0.0858,
+      "step": 7848
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.164077212806026e-05,
+      "loss": 0.1891,
+      "step": 7849
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1638418079096046e-05,
+      "loss": 0.0719,
+      "step": 7850
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.163606403013183e-05,
+      "loss": 0.1372,
+      "step": 7851
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.163370998116761e-05,
+      "loss": 0.1494,
+      "step": 7852
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.163135593220339e-05,
+      "loss": 0.1417,
+      "step": 7853
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1629001883239175e-05,
+      "loss": 0.214,
+      "step": 7854
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.162664783427495e-05,
+      "loss": 0.3863,
+      "step": 7855
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1624293785310736e-05,
+      "loss": 0.2171,
+      "step": 7856
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.162193973634652e-05,
+      "loss": 0.1603,
+      "step": 7857
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.16195856873823e-05,
+      "loss": 0.2114,
+      "step": 7858
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.161723163841808e-05,
+      "loss": 0.1338,
+      "step": 7859
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1614877589453865e-05,
+      "loss": 0.2013,
+      "step": 7860
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.161252354048964e-05,
+      "loss": 0.1586,
+      "step": 7861
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1610169491525426e-05,
+      "loss": 0.1113,
+      "step": 7862
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.16078154425612e-05,
+      "loss": 0.1821,
+      "step": 7863
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.160546139359699e-05,
+      "loss": 0.2012,
+      "step": 7864
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.160310734463277e-05,
+      "loss": 0.0633,
+      "step": 7865
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.160075329566855e-05,
+      "loss": 0.1892,
+      "step": 7866
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.159839924670433e-05,
+      "loss": 0.1001,
+      "step": 7867
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1596045197740116e-05,
+      "loss": 0.2521,
+      "step": 7868
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.159369114877589e-05,
+      "loss": 0.2527,
+      "step": 7869
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.159133709981168e-05,
+      "loss": 0.1382,
+      "step": 7870
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.158898305084746e-05,
+      "loss": 0.2321,
+      "step": 7871
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.158662900188324e-05,
+      "loss": 0.2102,
+      "step": 7872
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.158427495291902e-05,
+      "loss": 0.111,
+      "step": 7873
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1581920903954806e-05,
+      "loss": 0.1506,
+      "step": 7874
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.157956685499058e-05,
+      "loss": 0.2373,
+      "step": 7875
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.157721280602637e-05,
+      "loss": 0.1876,
+      "step": 7876
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.157485875706215e-05,
+      "loss": 0.1163,
+      "step": 7877
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.157250470809793e-05,
+      "loss": 0.0892,
+      "step": 7878
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.157015065913371e-05,
+      "loss": 0.1816,
+      "step": 7879
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.156779661016949e-05,
+      "loss": 0.1716,
+      "step": 7880
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.156544256120528e-05,
+      "loss": 0.21,
+      "step": 7881
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.156308851224106e-05,
+      "loss": 0.1444,
+      "step": 7882
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1560734463276834e-05,
+      "loss": 0.1409,
+      "step": 7883
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.155838041431262e-05,
+      "loss": 0.1612,
+      "step": 7884
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.15560263653484e-05,
+      "loss": 0.2922,
+      "step": 7885
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.155367231638418e-05,
+      "loss": 0.1213,
+      "step": 7886
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.155131826741996e-05,
+      "loss": 0.0898,
+      "step": 7887
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.154896421845575e-05,
+      "loss": 0.2172,
+      "step": 7888
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1546610169491524e-05,
+      "loss": 0.0793,
+      "step": 7889
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.154425612052731e-05,
+      "loss": 0.1955,
+      "step": 7890
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.154190207156309e-05,
+      "loss": 0.1326,
+      "step": 7891
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.153954802259887e-05,
+      "loss": 0.1646,
+      "step": 7892
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.153719397363465e-05,
+      "loss": 0.179,
+      "step": 7893
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.153483992467044e-05,
+      "loss": 0.1009,
+      "step": 7894
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1532485875706214e-05,
+      "loss": 0.091,
+      "step": 7895
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1530131826742e-05,
+      "loss": 0.1692,
+      "step": 7896
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1527777777777775e-05,
+      "loss": 0.2351,
+      "step": 7897
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1525423728813566e-05,
+      "loss": 0.044,
+      "step": 7898
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.152306967984934e-05,
+      "loss": 0.1704,
+      "step": 7899
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.152071563088512e-05,
+      "loss": 0.1036,
+      "step": 7900
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.151836158192091e-05,
+      "loss": 0.0588,
+      "step": 7901
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.151600753295669e-05,
+      "loss": 0.155,
+      "step": 7902
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1513653483992465e-05,
+      "loss": 0.1488,
+      "step": 7903
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.151129943502825e-05,
+      "loss": 0.0964,
+      "step": 7904
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.150894538606403e-05,
+      "loss": 0.1169,
+      "step": 7905
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.150659133709981e-05,
+      "loss": 0.1571,
+      "step": 7906
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1504237288135594e-05,
+      "loss": 0.0754,
+      "step": 7907
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.150188323917138e-05,
+      "loss": 0.2571,
+      "step": 7908
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1499529190207155e-05,
+      "loss": 0.1928,
+      "step": 7909
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.149717514124294e-05,
+      "loss": 0.1094,
+      "step": 7910
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.149482109227872e-05,
+      "loss": 0.1469,
+      "step": 7911
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.14924670433145e-05,
+      "loss": 0.1101,
+      "step": 7912
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1490112994350284e-05,
+      "loss": 0.225,
+      "step": 7913
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.148775894538606e-05,
+      "loss": 0.1133,
+      "step": 7914
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.148540489642185e-05,
+      "loss": 0.1065,
+      "step": 7915
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.148305084745763e-05,
+      "loss": 0.1762,
+      "step": 7916
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1480696798493406e-05,
+      "loss": 0.1106,
+      "step": 7917
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.14783427495292e-05,
+      "loss": 0.1183,
+      "step": 7918
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1475988700564974e-05,
+      "loss": 0.0634,
+      "step": 7919
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.147363465160075e-05,
+      "loss": 0.1478,
+      "step": 7920
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1471280602636535e-05,
+      "loss": 0.0851,
+      "step": 7921
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.146892655367232e-05,
+      "loss": 0.154,
+      "step": 7922
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1466572504708096e-05,
+      "loss": 0.1199,
+      "step": 7923
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.146421845574388e-05,
+      "loss": 0.2951,
+      "step": 7924
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1461864406779664e-05,
+      "loss": 0.194,
+      "step": 7925
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.145951035781544e-05,
+      "loss": 0.112,
+      "step": 7926
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1457156308851225e-05,
+      "loss": 0.0779,
+      "step": 7927
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.145480225988701e-05,
+      "loss": 0.1202,
+      "step": 7928
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.145244821092279e-05,
+      "loss": 0.1114,
+      "step": 7929
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.145009416195857e-05,
+      "loss": 0.1106,
+      "step": 7930
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.144774011299435e-05,
+      "loss": 0.1065,
+      "step": 7931
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.144538606403014e-05,
+      "loss": 0.1276,
+      "step": 7932
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1443032015065915e-05,
+      "loss": 0.2586,
+      "step": 7933
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.144067796610169e-05,
+      "loss": 0.1821,
+      "step": 7934
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.143832391713748e-05,
+      "loss": 0.1227,
+      "step": 7935
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.143596986817326e-05,
+      "loss": 0.123,
+      "step": 7936
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.143361581920904e-05,
+      "loss": 0.1375,
+      "step": 7937
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.143126177024482e-05,
+      "loss": 0.1966,
+      "step": 7938
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1428907721280605e-05,
+      "loss": 0.0582,
+      "step": 7939
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.142655367231638e-05,
+      "loss": 0.1229,
+      "step": 7940
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1424199623352166e-05,
+      "loss": 0.0264,
+      "step": 7941
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.142184557438795e-05,
+      "loss": 0.2179,
+      "step": 7942
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.141949152542373e-05,
+      "loss": 0.2307,
+      "step": 7943
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.141713747645951e-05,
+      "loss": 0.0184,
+      "step": 7944
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1414783427495295e-05,
+      "loss": 0.1278,
+      "step": 7945
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.141242937853108e-05,
+      "loss": 0.0976,
+      "step": 7946
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1410075329566856e-05,
+      "loss": 0.0815,
+      "step": 7947
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1407721280602634e-05,
+      "loss": 0.2146,
+      "step": 7948
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1405367231638424e-05,
+      "loss": 0.1489,
+      "step": 7949
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.14030131826742e-05,
+      "loss": 0.206,
+      "step": 7950
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.140065913370998e-05,
+      "loss": 0.0513,
+      "step": 7951
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.139830508474577e-05,
+      "loss": 0.2586,
+      "step": 7952
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1395951035781546e-05,
+      "loss": 0.1567,
+      "step": 7953
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1393596986817324e-05,
+      "loss": 0.0992,
+      "step": 7954
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.139124293785311e-05,
+      "loss": 0.1159,
+      "step": 7955
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.138888888888889e-05,
+      "loss": 0.1257,
+      "step": 7956
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.138653483992467e-05,
+      "loss": 0.1183,
+      "step": 7957
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.138418079096045e-05,
+      "loss": 0.0762,
+      "step": 7958
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1381826741996236e-05,
+      "loss": 0.2766,
+      "step": 7959
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.137947269303202e-05,
+      "loss": 0.1358,
+      "step": 7960
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.13771186440678e-05,
+      "loss": 0.0832,
+      "step": 7961
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.137476459510358e-05,
+      "loss": 0.2199,
+      "step": 7962
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1372410546139365e-05,
+      "loss": 0.0441,
+      "step": 7963
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.137005649717514e-05,
+      "loss": 0.1767,
+      "step": 7964
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.136770244821092e-05,
+      "loss": 0.0914,
+      "step": 7965
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.136534839924671e-05,
+      "loss": 0.1102,
+      "step": 7966
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.136299435028249e-05,
+      "loss": 0.3309,
+      "step": 7967
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1360640301318265e-05,
+      "loss": 0.0722,
+      "step": 7968
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1358286252354055e-05,
+      "loss": 0.0653,
+      "step": 7969
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.135593220338983e-05,
+      "loss": 0.1906,
+      "step": 7970
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.135357815442561e-05,
+      "loss": 0.297,
+      "step": 7971
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1351224105461394e-05,
+      "loss": 0.1993,
+      "step": 7972
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.134887005649718e-05,
+      "loss": 0.094,
+      "step": 7973
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1346516007532955e-05,
+      "loss": 0.0894,
+      "step": 7974
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.134416195856874e-05,
+      "loss": 0.0692,
+      "step": 7975
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.134180790960452e-05,
+      "loss": 0.0559,
+      "step": 7976
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1339453860640306e-05,
+      "loss": 0.2284,
+      "step": 7977
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1337099811676084e-05,
+      "loss": 0.0934,
+      "step": 7978
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.133474576271187e-05,
+      "loss": 0.0396,
+      "step": 7979
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.133239171374765e-05,
+      "loss": 0.1738,
+      "step": 7980
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.133003766478343e-05,
+      "loss": 0.1231,
+      "step": 7981
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1327683615819206e-05,
+      "loss": 0.1214,
+      "step": 7982
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1325329566854996e-05,
+      "loss": 0.1069,
+      "step": 7983
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1322975517890774e-05,
+      "loss": 0.1459,
+      "step": 7984
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.132062146892655e-05,
+      "loss": 0.1755,
+      "step": 7985
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.131826741996234e-05,
+      "loss": 0.1945,
+      "step": 7986
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.131591337099812e-05,
+      "loss": 0.1357,
+      "step": 7987
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1313559322033896e-05,
+      "loss": 0.2143,
+      "step": 7988
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.131120527306968e-05,
+      "loss": 0.1531,
+      "step": 7989
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1308851224105464e-05,
+      "loss": 0.0733,
+      "step": 7990
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.130649717514124e-05,
+      "loss": 0.1067,
+      "step": 7991
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1304143126177025e-05,
+      "loss": 0.1168,
+      "step": 7992
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.130178907721281e-05,
+      "loss": 0.1164,
+      "step": 7993
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.129943502824859e-05,
+      "loss": 0.0838,
+      "step": 7994
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.129708097928437e-05,
+      "loss": 0.1092,
+      "step": 7995
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1294726930320154e-05,
+      "loss": 0.1032,
+      "step": 7996
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.129237288135594e-05,
+      "loss": 0.2442,
+      "step": 7997
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1290018832391715e-05,
+      "loss": 0.2478,
+      "step": 7998
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.128766478342749e-05,
+      "loss": 0.0714,
+      "step": 7999
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.128531073446328e-05,
+      "loss": 0.1795,
+      "step": 8000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.128295668549906e-05,
+      "loss": 0.082,
+      "step": 8001
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.128060263653484e-05,
+      "loss": 0.122,
+      "step": 8002
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.127824858757063e-05,
+      "loss": 0.0522,
+      "step": 8003
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1275894538606405e-05,
+      "loss": 0.0624,
+      "step": 8004
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.127354048964218e-05,
+      "loss": 0.0685,
+      "step": 8005
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1271186440677966e-05,
+      "loss": 0.0997,
+      "step": 8006
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.126883239171375e-05,
+      "loss": 0.1627,
+      "step": 8007
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1266478342749534e-05,
+      "loss": 0.2031,
+      "step": 8008
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.126412429378531e-05,
+      "loss": 0.2084,
+      "step": 8009
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1261770244821095e-05,
+      "loss": 0.0882,
+      "step": 8010
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.125941619585688e-05,
+      "loss": 0.0938,
+      "step": 8011
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1257062146892656e-05,
+      "loss": 0.1404,
+      "step": 8012
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.125470809792844e-05,
+      "loss": 0.131,
+      "step": 8013
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1252354048964224e-05,
+      "loss": 0.1701,
+      "step": 8014
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.125e-05,
+      "loss": 0.0667,
+      "step": 8015
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.124764595103578e-05,
+      "loss": 0.0375,
+      "step": 8016
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.124529190207157e-05,
+      "loss": 0.1225,
+      "step": 8017
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1242937853107346e-05,
+      "loss": 0.0705,
+      "step": 8018
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.124058380414312e-05,
+      "loss": 0.1205,
+      "step": 8019
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1238229755178913e-05,
+      "loss": 0.0763,
+      "step": 8020
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.123587570621469e-05,
+      "loss": 0.0474,
+      "step": 8021
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.123352165725047e-05,
+      "loss": 0.2207,
+      "step": 8022
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.123116760828625e-05,
+      "loss": 0.0743,
+      "step": 8023
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1228813559322036e-05,
+      "loss": 0.1576,
+      "step": 8024
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.122645951035782e-05,
+      "loss": 0.0859,
+      "step": 8025
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.12241054613936e-05,
+      "loss": 0.12,
+      "step": 8026
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.122175141242938e-05,
+      "loss": 0.1122,
+      "step": 8027
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1219397363465165e-05,
+      "loss": 0.0833,
+      "step": 8028
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.121704331450094e-05,
+      "loss": 0.1397,
+      "step": 8029
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1214689265536726e-05,
+      "loss": 0.0665,
+      "step": 8030
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.121233521657251e-05,
+      "loss": 0.1224,
+      "step": 8031
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.120998116760829e-05,
+      "loss": 0.1375,
+      "step": 8032
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1207627118644064e-05,
+      "loss": 0.2748,
+      "step": 8033
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1205273069679855e-05,
+      "loss": 0.0393,
+      "step": 8034
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.120291902071563e-05,
+      "loss": 0.0832,
+      "step": 8035
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.120056497175141e-05,
+      "loss": 0.142,
+      "step": 8036
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.11982109227872e-05,
+      "loss": 0.2844,
+      "step": 8037
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.119585687382298e-05,
+      "loss": 0.0625,
+      "step": 8038
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.119350282485876e-05,
+      "loss": 0.1321,
+      "step": 8039
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.119114877589454e-05,
+      "loss": 0.1371,
+      "step": 8040
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.118879472693032e-05,
+      "loss": 0.0481,
+      "step": 8041
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1186440677966106e-05,
+      "loss": 0.116,
+      "step": 8042
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.118408662900188e-05,
+      "loss": 0.1424,
+      "step": 8043
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.118173258003767e-05,
+      "loss": 0.1825,
+      "step": 8044
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.117937853107345e-05,
+      "loss": 0.2448,
+      "step": 8045
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.117702448210923e-05,
+      "loss": 0.1527,
+      "step": 8046
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.117467043314501e-05,
+      "loss": 0.1546,
+      "step": 8047
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1172316384180796e-05,
+      "loss": 0.1023,
+      "step": 8048
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.116996233521657e-05,
+      "loss": 0.1917,
+      "step": 8049
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.116760828625235e-05,
+      "loss": 0.0648,
+      "step": 8050
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.116525423728814e-05,
+      "loss": 0.0902,
+      "step": 8051
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.116290018832392e-05,
+      "loss": 0.1834,
+      "step": 8052
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1160546139359695e-05,
+      "loss": 0.1105,
+      "step": 8053
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1158192090395486e-05,
+      "loss": 0.1559,
+      "step": 8054
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.115583804143126e-05,
+      "loss": 0.1117,
+      "step": 8055
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.115348399246705e-05,
+      "loss": 0.1889,
+      "step": 8056
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1151129943502824e-05,
+      "loss": 0.1766,
+      "step": 8057
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.114877589453861e-05,
+      "loss": 0.2312,
+      "step": 8058
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.114642184557439e-05,
+      "loss": 0.0839,
+      "step": 8059
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.114406779661017e-05,
+      "loss": 0.0766,
+      "step": 8060
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.114171374764595e-05,
+      "loss": 0.1409,
+      "step": 8061
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.113935969868174e-05,
+      "loss": 0.0452,
+      "step": 8062
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1137005649717514e-05,
+      "loss": 0.092,
+      "step": 8063
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.11346516007533e-05,
+      "loss": 0.1474,
+      "step": 8064
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.113229755178908e-05,
+      "loss": 0.0771,
+      "step": 8065
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.112994350282486e-05,
+      "loss": 0.0939,
+      "step": 8066
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.112758945386064e-05,
+      "loss": 0.1357,
+      "step": 8067
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.112523540489643e-05,
+      "loss": 0.1942,
+      "step": 8068
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1122881355932204e-05,
+      "loss": 0.3025,
+      "step": 8069
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.112052730696798e-05,
+      "loss": 0.127,
+      "step": 8070
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.111817325800377e-05,
+      "loss": 0.1682,
+      "step": 8071
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.111581920903955e-05,
+      "loss": 0.0964,
+      "step": 8072
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.111346516007533e-05,
+      "loss": 0.1862,
+      "step": 8073
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.111111111111111e-05,
+      "loss": 0.2143,
+      "step": 8074
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1108757062146894e-05,
+      "loss": 0.2027,
+      "step": 8075
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.110640301318268e-05,
+      "loss": 0.0572,
+      "step": 8076
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1104048964218455e-05,
+      "loss": 0.1459,
+      "step": 8077
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.110169491525424e-05,
+      "loss": 0.075,
+      "step": 8078
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.109934086629002e-05,
+      "loss": 0.2496,
+      "step": 8079
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.10969868173258e-05,
+      "loss": 0.1259,
+      "step": 8080
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1094632768361584e-05,
+      "loss": 0.1541,
+      "step": 8081
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.109227871939737e-05,
+      "loss": 0.1329,
+      "step": 8082
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1089924670433145e-05,
+      "loss": 0.1162,
+      "step": 8083
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.108757062146893e-05,
+      "loss": 0.3019,
+      "step": 8084
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.108521657250471e-05,
+      "loss": 0.125,
+      "step": 8085
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.108286252354049e-05,
+      "loss": 0.1465,
+      "step": 8086
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1080508474576274e-05,
+      "loss": 0.1191,
+      "step": 8087
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.107815442561206e-05,
+      "loss": 0.1419,
+      "step": 8088
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1075800376647835e-05,
+      "loss": 0.0482,
+      "step": 8089
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.107344632768362e-05,
+      "loss": 0.1181,
+      "step": 8090
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1071092278719396e-05,
+      "loss": 0.1926,
+      "step": 8091
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.106873822975518e-05,
+      "loss": 0.0908,
+      "step": 8092
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1066384180790964e-05,
+      "loss": 0.0402,
+      "step": 8093
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.106403013182674e-05,
+      "loss": 0.0885,
+      "step": 8094
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1061676082862525e-05,
+      "loss": 0.0786,
+      "step": 8095
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.105932203389831e-05,
+      "loss": 0.1541,
+      "step": 8096
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1056967984934086e-05,
+      "loss": 0.0366,
+      "step": 8097
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.105461393596987e-05,
+      "loss": 0.1135,
+      "step": 8098
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1052259887005654e-05,
+      "loss": 0.1637,
+      "step": 8099
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.104990583804143e-05,
+      "loss": 0.2402,
+      "step": 8100
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.1047551789077215e-05,
+      "loss": 0.1798,
+      "step": 8101
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.1045197740113e-05,
+      "loss": 0.1251,
+      "step": 8102
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.1042843691148776e-05,
+      "loss": 0.0426,
+      "step": 8103
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.104048964218456e-05,
+      "loss": 0.0985,
+      "step": 8104
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.1038135593220344e-05,
+      "loss": 0.0912,
+      "step": 8105
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.103578154425612e-05,
+      "loss": 0.1177,
+      "step": 8106
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.1033427495291905e-05,
+      "loss": 0.1862,
+      "step": 8107
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.103107344632768e-05,
+      "loss": 0.114,
+      "step": 8108
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.1028719397363466e-05,
+      "loss": 0.2483,
+      "step": 8109
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.102636534839925e-05,
+      "loss": 0.195,
+      "step": 8110
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.102401129943503e-05,
+      "loss": 0.1447,
+      "step": 8111
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.102165725047081e-05,
+      "loss": 0.1095,
+      "step": 8112
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.1019303201506595e-05,
+      "loss": 0.2919,
+      "step": 8113
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.101694915254237e-05,
+      "loss": 0.2117,
+      "step": 8114
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.1014595103578156e-05,
+      "loss": 0.2097,
+      "step": 8115
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.101224105461394e-05,
+      "loss": 0.0476,
+      "step": 8116
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.100988700564972e-05,
+      "loss": 0.0335,
+      "step": 8117
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.10075329566855e-05,
+      "loss": 0.1659,
+      "step": 8118
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.1005178907721285e-05,
+      "loss": 0.1052,
+      "step": 8119
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.100282485875706e-05,
+      "loss": 0.1136,
+      "step": 8120
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.1000470809792846e-05,
+      "loss": 0.182,
+      "step": 8121
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.099811676082863e-05,
+      "loss": 0.1431,
+      "step": 8122
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.099576271186441e-05,
+      "loss": 0.0597,
+      "step": 8123
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.099340866290019e-05,
+      "loss": 0.2136,
+      "step": 8124
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.099105461393597e-05,
+      "loss": 0.077,
+      "step": 8125
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.098870056497175e-05,
+      "loss": 0.0884,
+      "step": 8126
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0986346516007536e-05,
+      "loss": 0.1572,
+      "step": 8127
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.098399246704331e-05,
+      "loss": 0.1435,
+      "step": 8128
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.09816384180791e-05,
+      "loss": 0.2159,
+      "step": 8129
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.097928436911488e-05,
+      "loss": 0.2292,
+      "step": 8130
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.097693032015066e-05,
+      "loss": 0.0894,
+      "step": 8131
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.097457627118644e-05,
+      "loss": 0.1266,
+      "step": 8132
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0972222222222226e-05,
+      "loss": 0.0896,
+      "step": 8133
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0969868173258e-05,
+      "loss": 0.072,
+      "step": 8134
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.096751412429379e-05,
+      "loss": 0.0861,
+      "step": 8135
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.096516007532957e-05,
+      "loss": 0.1566,
+      "step": 8136
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.096280602636535e-05,
+      "loss": 0.0275,
+      "step": 8137
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.096045197740113e-05,
+      "loss": 0.15,
+      "step": 8138
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0958097928436916e-05,
+      "loss": 0.1142,
+      "step": 8139
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.095574387947269e-05,
+      "loss": 0.1102,
+      "step": 8140
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.095338983050848e-05,
+      "loss": 0.1673,
+      "step": 8141
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.095103578154426e-05,
+      "loss": 0.0793,
+      "step": 8142
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.094868173258004e-05,
+      "loss": 0.109,
+      "step": 8143
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.094632768361582e-05,
+      "loss": 0.1902,
+      "step": 8144
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.09439736346516e-05,
+      "loss": 0.2344,
+      "step": 8145
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.094161958568738e-05,
+      "loss": 0.0655,
+      "step": 8146
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.093926553672317e-05,
+      "loss": 0.0738,
+      "step": 8147
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0936911487758944e-05,
+      "loss": 0.1253,
+      "step": 8148
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.093455743879473e-05,
+      "loss": 0.113,
+      "step": 8149
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.093220338983051e-05,
+      "loss": 0.0488,
+      "step": 8150
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.092984934086629e-05,
+      "loss": 0.1749,
+      "step": 8151
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.092749529190207e-05,
+      "loss": 0.1401,
+      "step": 8152
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.092514124293786e-05,
+      "loss": 0.0953,
+      "step": 8153
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0922787193973634e-05,
+      "loss": 0.0851,
+      "step": 8154
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.092043314500942e-05,
+      "loss": 0.2885,
+      "step": 8155
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.09180790960452e-05,
+      "loss": 0.0591,
+      "step": 8156
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.091572504708098e-05,
+      "loss": 0.0698,
+      "step": 8157
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.091337099811676e-05,
+      "loss": 0.2019,
+      "step": 8158
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.091101694915255e-05,
+      "loss": 0.1515,
+      "step": 8159
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0908662900188324e-05,
+      "loss": 0.0999,
+      "step": 8160
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.090630885122411e-05,
+      "loss": 0.1128,
+      "step": 8161
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0903954802259885e-05,
+      "loss": 0.1711,
+      "step": 8162
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.090160075329567e-05,
+      "loss": 0.14,
+      "step": 8163
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.089924670433145e-05,
+      "loss": 0.1531,
+      "step": 8164
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.089689265536723e-05,
+      "loss": 0.1437,
+      "step": 8165
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0894538606403014e-05,
+      "loss": 0.1971,
+      "step": 8166
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.08921845574388e-05,
+      "loss": 0.1356,
+      "step": 8167
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0889830508474575e-05,
+      "loss": 0.0453,
+      "step": 8168
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.088747645951036e-05,
+      "loss": 0.07,
+      "step": 8169
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.088512241054614e-05,
+      "loss": 0.1734,
+      "step": 8170
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.088276836158192e-05,
+      "loss": 0.0692,
+      "step": 8171
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0880414312617704e-05,
+      "loss": 0.175,
+      "step": 8172
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.087806026365349e-05,
+      "loss": 0.0405,
+      "step": 8173
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0875706214689265e-05,
+      "loss": 0.1071,
+      "step": 8174
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.087335216572505e-05,
+      "loss": 0.0169,
+      "step": 8175
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.087099811676083e-05,
+      "loss": 0.091,
+      "step": 8176
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.086864406779661e-05,
+      "loss": 0.1039,
+      "step": 8177
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0866290018832394e-05,
+      "loss": 0.1138,
+      "step": 8178
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.086393596986817e-05,
+      "loss": 0.1652,
+      "step": 8179
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0861581920903955e-05,
+      "loss": 0.1699,
+      "step": 8180
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.085922787193974e-05,
+      "loss": 0.0658,
+      "step": 8181
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0856873822975516e-05,
+      "loss": 0.2544,
+      "step": 8182
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.08545197740113e-05,
+      "loss": 0.2366,
+      "step": 8183
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0852165725047084e-05,
+      "loss": 0.1339,
+      "step": 8184
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.084981167608286e-05,
+      "loss": 0.2237,
+      "step": 8185
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0847457627118645e-05,
+      "loss": 0.136,
+      "step": 8186
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.084510357815443e-05,
+      "loss": 0.114,
+      "step": 8187
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0842749529190206e-05,
+      "loss": 0.1053,
+      "step": 8188
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.084039548022599e-05,
+      "loss": 0.2621,
+      "step": 8189
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0838041431261774e-05,
+      "loss": 0.1251,
+      "step": 8190
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.083568738229755e-05,
+      "loss": 0.0968,
+      "step": 8191
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0833333333333335e-05,
+      "loss": 0.1276,
+      "step": 8192
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.083097928436912e-05,
+      "loss": 0.1527,
+      "step": 8193
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0828625235404896e-05,
+      "loss": 0.2137,
+      "step": 8194
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.082627118644068e-05,
+      "loss": 0.2363,
+      "step": 8195
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.082391713747646e-05,
+      "loss": 0.1364,
+      "step": 8196
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.082156308851225e-05,
+      "loss": 0.0551,
+      "step": 8197
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0819209039548025e-05,
+      "loss": 0.1498,
+      "step": 8198
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.08168549905838e-05,
+      "loss": 0.0624,
+      "step": 8199
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0814500941619586e-05,
+      "loss": 0.1072,
+      "step": 8200
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.081214689265537e-05,
+      "loss": 0.2947,
+      "step": 8201
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.080979284369115e-05,
+      "loss": 0.1134,
+      "step": 8202
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.080743879472693e-05,
+      "loss": 0.0676,
+      "step": 8203
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0805084745762715e-05,
+      "loss": 0.0981,
+      "step": 8204
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.080273069679849e-05,
+      "loss": 0.0185,
+      "step": 8205
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0800376647834276e-05,
+      "loss": 0.1303,
+      "step": 8206
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.079802259887006e-05,
+      "loss": 0.1631,
+      "step": 8207
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.079566854990584e-05,
+      "loss": 0.1069,
+      "step": 8208
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.079331450094162e-05,
+      "loss": 0.0482,
+      "step": 8209
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0790960451977405e-05,
+      "loss": 0.1874,
+      "step": 8210
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.078860640301318e-05,
+      "loss": 0.1606,
+      "step": 8211
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0786252354048966e-05,
+      "loss": 0.3201,
+      "step": 8212
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.078389830508474e-05,
+      "loss": 0.0772,
+      "step": 8213
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0781544256120534e-05,
+      "loss": 0.0783,
+      "step": 8214
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.077919020715631e-05,
+      "loss": 0.2533,
+      "step": 8215
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.077683615819209e-05,
+      "loss": 0.1291,
+      "step": 8216
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.077448210922788e-05,
+      "loss": 0.1149,
+      "step": 8217
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0772128060263656e-05,
+      "loss": 0.1521,
+      "step": 8218
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.076977401129943e-05,
+      "loss": 0.1236,
+      "step": 8219
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.076741996233522e-05,
+      "loss": 0.1939,
+      "step": 8220
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0765065913371e-05,
+      "loss": 0.0944,
+      "step": 8221
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.076271186440678e-05,
+      "loss": 0.2261,
+      "step": 8222
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.076035781544256e-05,
+      "loss": 0.1589,
+      "step": 8223
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0758003766478346e-05,
+      "loss": 0.189,
+      "step": 8224
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.075564971751412e-05,
+      "loss": 0.0806,
+      "step": 8225
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.075329566854991e-05,
+      "loss": 0.0856,
+      "step": 8226
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.075094161958569e-05,
+      "loss": 0.1329,
+      "step": 8227
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.074858757062147e-05,
+      "loss": 0.2296,
+      "step": 8228
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.074623352165725e-05,
+      "loss": 0.2076,
+      "step": 8229
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.074387947269303e-05,
+      "loss": 0.2114,
+      "step": 8230
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.074152542372882e-05,
+      "loss": 0.1727,
+      "step": 8231
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.07391713747646e-05,
+      "loss": 0.1063,
+      "step": 8232
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0736817325800374e-05,
+      "loss": 0.0962,
+      "step": 8233
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0734463276836165e-05,
+      "loss": 0.1823,
+      "step": 8234
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.073210922787194e-05,
+      "loss": 0.0995,
+      "step": 8235
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.072975517890772e-05,
+      "loss": 0.1535,
+      "step": 8236
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.07274011299435e-05,
+      "loss": 0.2795,
+      "step": 8237
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.072504708097929e-05,
+      "loss": 0.1697,
+      "step": 8238
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0722693032015064e-05,
+      "loss": 0.0727,
+      "step": 8239
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.072033898305085e-05,
+      "loss": 0.2411,
+      "step": 8240
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.071798493408663e-05,
+      "loss": 0.1819,
+      "step": 8241
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.071563088512241e-05,
+      "loss": 0.0984,
+      "step": 8242
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.071327683615819e-05,
+      "loss": 0.2236,
+      "step": 8243
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.071092278719398e-05,
+      "loss": 0.1093,
+      "step": 8244
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.070856873822976e-05,
+      "loss": 0.1018,
+      "step": 8245
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.070621468926554e-05,
+      "loss": 0.0748,
+      "step": 8246
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0703860640301315e-05,
+      "loss": 0.1458,
+      "step": 8247
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0701506591337106e-05,
+      "loss": 0.1016,
+      "step": 8248
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.069915254237288e-05,
+      "loss": 0.2007,
+      "step": 8249
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.069679849340866e-05,
+      "loss": 0.0926,
+      "step": 8250
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.069444444444445e-05,
+      "loss": 0.096,
+      "step": 8251
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.069209039548023e-05,
+      "loss": 0.1348,
+      "step": 8252
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0689736346516005e-05,
+      "loss": 0.2496,
+      "step": 8253
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.068738229755179e-05,
+      "loss": 0.2233,
+      "step": 8254
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.068502824858757e-05,
+      "loss": 0.205,
+      "step": 8255
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.068267419962335e-05,
+      "loss": 0.1583,
+      "step": 8256
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0680320150659134e-05,
+      "loss": 0.1695,
+      "step": 8257
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.067796610169492e-05,
+      "loss": 0.0892,
+      "step": 8258
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0675612052730695e-05,
+      "loss": 0.1827,
+      "step": 8259
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.067325800376648e-05,
+      "loss": 0.1513,
+      "step": 8260
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.067090395480226e-05,
+      "loss": 0.1608,
+      "step": 8261
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.066854990583805e-05,
+      "loss": 0.1023,
+      "step": 8262
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0666195856873824e-05,
+      "loss": 0.1504,
+      "step": 8263
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.06638418079096e-05,
+      "loss": 0.0713,
+      "step": 8264
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.066148775894539e-05,
+      "loss": 0.1301,
+      "step": 8265
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.065913370998117e-05,
+      "loss": 0.1499,
+      "step": 8266
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0656779661016947e-05,
+      "loss": 0.1217,
+      "step": 8267
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.065442561205274e-05,
+      "loss": 0.1704,
+      "step": 8268
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0652071563088514e-05,
+      "loss": 0.1882,
+      "step": 8269
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.064971751412429e-05,
+      "loss": 0.1099,
+      "step": 8270
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0647363465160075e-05,
+      "loss": 0.117,
+      "step": 8271
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.064500941619586e-05,
+      "loss": 0.1508,
+      "step": 8272
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0642655367231637e-05,
+      "loss": 0.1181,
+      "step": 8273
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.064030131826742e-05,
+      "loss": 0.1023,
+      "step": 8274
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0637947269303204e-05,
+      "loss": 0.1359,
+      "step": 8275
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.063559322033899e-05,
+      "loss": 0.1395,
+      "step": 8276
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0633239171374765e-05,
+      "loss": 0.1131,
+      "step": 8277
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.063088512241055e-05,
+      "loss": 0.1141,
+      "step": 8278
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.062853107344633e-05,
+      "loss": 0.1022,
+      "step": 8279
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.062617702448211e-05,
+      "loss": 0.0979,
+      "step": 8280
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.062382297551789e-05,
+      "loss": 0.1613,
+      "step": 8281
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.062146892655368e-05,
+      "loss": 0.0637,
+      "step": 8282
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0619114877589455e-05,
+      "loss": 0.2049,
+      "step": 8283
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.061676082862523e-05,
+      "loss": 0.0805,
+      "step": 8284
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.061440677966102e-05,
+      "loss": 0.072,
+      "step": 8285
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.06120527306968e-05,
+      "loss": 0.1565,
+      "step": 8286
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.060969868173258e-05,
+      "loss": 0.258,
+      "step": 8287
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.060734463276836e-05,
+      "loss": 0.1156,
+      "step": 8288
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0604990583804145e-05,
+      "loss": 0.1684,
+      "step": 8289
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.060263653483992e-05,
+      "loss": 0.1226,
+      "step": 8290
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0600282485875707e-05,
+      "loss": 0.1726,
+      "step": 8291
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.059792843691149e-05,
+      "loss": 0.1528,
+      "step": 8292
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0595574387947274e-05,
+      "loss": 0.1424,
+      "step": 8293
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.059322033898305e-05,
+      "loss": 0.1715,
+      "step": 8294
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0590866290018835e-05,
+      "loss": 0.0909,
+      "step": 8295
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.058851224105462e-05,
+      "loss": 0.1624,
+      "step": 8296
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0586158192090396e-05,
+      "loss": 0.0347,
+      "step": 8297
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0583804143126174e-05,
+      "loss": 0.1191,
+      "step": 8298
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0581450094161964e-05,
+      "loss": 0.0589,
+      "step": 8299
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.057909604519774e-05,
+      "loss": 0.142,
+      "step": 8300
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.057674199623352e-05,
+      "loss": 0.2207,
+      "step": 8301
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.057438794726931e-05,
+      "loss": 0.0718,
+      "step": 8302
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0572033898305086e-05,
+      "loss": 0.2954,
+      "step": 8303
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0569679849340864e-05,
+      "loss": 0.1549,
+      "step": 8304
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.056732580037665e-05,
+      "loss": 0.1695,
+      "step": 8305
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.056497175141243e-05,
+      "loss": 0.2229,
+      "step": 8306
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.056261770244821e-05,
+      "loss": 0.0808,
+      "step": 8307
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.056026365348399e-05,
+      "loss": 0.0917,
+      "step": 8308
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0557909604519776e-05,
+      "loss": 0.1464,
+      "step": 8309
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.055555555555556e-05,
+      "loss": 0.1505,
+      "step": 8310
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.055320150659134e-05,
+      "loss": 0.1025,
+      "step": 8311
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.055084745762712e-05,
+      "loss": 0.0934,
+      "step": 8312
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0548493408662905e-05,
+      "loss": 0.0888,
+      "step": 8313
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.054613935969868e-05,
+      "loss": 0.0407,
+      "step": 8314
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.054378531073446e-05,
+      "loss": 0.1117,
+      "step": 8315
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.054143126177025e-05,
+      "loss": 0.1296,
+      "step": 8316
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.053907721280603e-05,
+      "loss": 0.0506,
+      "step": 8317
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0536723163841805e-05,
+      "loss": 0.1154,
+      "step": 8318
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0534369114877595e-05,
+      "loss": 0.1096,
+      "step": 8319
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.053201506591337e-05,
+      "loss": 0.0748,
+      "step": 8320
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.052966101694915e-05,
+      "loss": 0.0671,
+      "step": 8321
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0527306967984934e-05,
+      "loss": 0.1203,
+      "step": 8322
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.052495291902072e-05,
+      "loss": 0.2042,
+      "step": 8323
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.05225988700565e-05,
+      "loss": 0.104,
+      "step": 8324
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.052024482109228e-05,
+      "loss": 0.2006,
+      "step": 8325
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.051789077212806e-05,
+      "loss": 0.1777,
+      "step": 8326
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0515536723163846e-05,
+      "loss": 0.1358,
+      "step": 8327
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0513182674199624e-05,
+      "loss": 0.148,
+      "step": 8328
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0510828625235404e-05,
+      "loss": 0.1329,
+      "step": 8329
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.050847457627119e-05,
+      "loss": 0.2007,
+      "step": 8330
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.050612052730697e-05,
+      "loss": 0.1554,
+      "step": 8331
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.050376647834275e-05,
+      "loss": 0.1601,
+      "step": 8332
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0501412429378533e-05,
+      "loss": 0.3139,
+      "step": 8333
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0499058380414314e-05,
+      "loss": 0.1831,
+      "step": 8334
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0496704331450094e-05,
+      "loss": 0.1986,
+      "step": 8335
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0494350282485878e-05,
+      "loss": 0.0904,
+      "step": 8336
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.049199623352166e-05,
+      "loss": 0.1176,
+      "step": 8337
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.048964218455744e-05,
+      "loss": 0.0707,
+      "step": 8338
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0487288135593223e-05,
+      "loss": 0.1215,
+      "step": 8339
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0484934086629004e-05,
+      "loss": 0.1383,
+      "step": 8340
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0482580037664788e-05,
+      "loss": 0.0997,
+      "step": 8341
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0480225988700568e-05,
+      "loss": 0.1878,
+      "step": 8342
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0477871939736345e-05,
+      "loss": 0.078,
+      "step": 8343
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0475517890772133e-05,
+      "loss": 0.2026,
+      "step": 8344
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.047316384180791e-05,
+      "loss": 0.1019,
+      "step": 8345
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.047080979284369e-05,
+      "loss": 0.1142,
+      "step": 8346
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0468455743879478e-05,
+      "loss": 0.2498,
+      "step": 8347
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0466101694915255e-05,
+      "loss": 0.0796,
+      "step": 8348
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0463747645951035e-05,
+      "loss": 0.0808,
+      "step": 8349
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.046139359698682e-05,
+      "loss": 0.1636,
+      "step": 8350
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.04590395480226e-05,
+      "loss": 0.2271,
+      "step": 8351
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.045668549905838e-05,
+      "loss": 0.2091,
+      "step": 8352
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0454331450094164e-05,
+      "loss": 0.1213,
+      "step": 8353
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0451977401129945e-05,
+      "loss": 0.1827,
+      "step": 8354
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.044962335216573e-05,
+      "loss": 0.1396,
+      "step": 8355
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.044726930320151e-05,
+      "loss": 0.1134,
+      "step": 8356
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.044491525423729e-05,
+      "loss": 0.1212,
+      "step": 8357
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0442561205273074e-05,
+      "loss": 0.1592,
+      "step": 8358
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0440207156308854e-05,
+      "loss": 0.116,
+      "step": 8359
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.043785310734463e-05,
+      "loss": 0.0914,
+      "step": 8360
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.043549905838042e-05,
+      "loss": 0.0737,
+      "step": 8361
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0433145009416196e-05,
+      "loss": 0.2156,
+      "step": 8362
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0430790960451976e-05,
+      "loss": 0.1785,
+      "step": 8363
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0428436911487764e-05,
+      "loss": 0.1116,
+      "step": 8364
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.042608286252354e-05,
+      "loss": 0.095,
+      "step": 8365
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.042372881355932e-05,
+      "loss": 0.0861,
+      "step": 8366
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0421374764595105e-05,
+      "loss": 0.1668,
+      "step": 8367
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0419020715630886e-05,
+      "loss": 0.1224,
+      "step": 8368
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0416666666666666e-05,
+      "loss": 0.0754,
+      "step": 8369
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.041431261770245e-05,
+      "loss": 0.1255,
+      "step": 8370
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.041195856873823e-05,
+      "loss": 0.0783,
+      "step": 8371
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0409604519774015e-05,
+      "loss": 0.1644,
+      "step": 8372
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0407250470809795e-05,
+      "loss": 0.074,
+      "step": 8373
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0404896421845576e-05,
+      "loss": 0.0697,
+      "step": 8374
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.040254237288136e-05,
+      "loss": 0.195,
+      "step": 8375
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.040018832391714e-05,
+      "loss": 0.141,
+      "step": 8376
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0397834274952917e-05,
+      "loss": 0.2064,
+      "step": 8377
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0395480225988705e-05,
+      "loss": 0.1305,
+      "step": 8378
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0393126177024485e-05,
+      "loss": 0.1607,
+      "step": 8379
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0390772128060262e-05,
+      "loss": 0.1106,
+      "step": 8380
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.038841807909605e-05,
+      "loss": 0.0691,
+      "step": 8381
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0386064030131827e-05,
+      "loss": 0.119,
+      "step": 8382
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0383709981167607e-05,
+      "loss": 0.1102,
+      "step": 8383
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.038135593220339e-05,
+      "loss": 0.1643,
+      "step": 8384
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0379001883239172e-05,
+      "loss": 0.1777,
+      "step": 8385
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0376647834274952e-05,
+      "loss": 0.1314,
+      "step": 8386
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0374293785310736e-05,
+      "loss": 0.1025,
+      "step": 8387
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0371939736346517e-05,
+      "loss": 0.1222,
+      "step": 8388
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.03695856873823e-05,
+      "loss": 0.1383,
+      "step": 8389
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.036723163841808e-05,
+      "loss": 0.1823,
+      "step": 8390
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0364877589453862e-05,
+      "loss": 0.1405,
+      "step": 8391
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0362523540489646e-05,
+      "loss": 0.109,
+      "step": 8392
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0360169491525426e-05,
+      "loss": 0.073,
+      "step": 8393
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0357815442561203e-05,
+      "loss": 0.1126,
+      "step": 8394
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.035546139359699e-05,
+      "loss": 0.15,
+      "step": 8395
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.035310734463277e-05,
+      "loss": 0.1501,
+      "step": 8396
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.035075329566855e-05,
+      "loss": 0.109,
+      "step": 8397
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0348399246704336e-05,
+      "loss": 0.1284,
+      "step": 8398
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0346045197740113e-05,
+      "loss": 0.1086,
+      "step": 8399
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0343691148775893e-05,
+      "loss": 0.1712,
+      "step": 8400
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0341337099811677e-05,
+      "loss": 0.1623,
+      "step": 8401
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0338983050847458e-05,
+      "loss": 0.2636,
+      "step": 8402
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0336629001883242e-05,
+      "loss": 0.1823,
+      "step": 8403
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0334274952919022e-05,
+      "loss": 0.3143,
+      "step": 8404
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0331920903954803e-05,
+      "loss": 0.1941,
+      "step": 8405
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0329566854990587e-05,
+      "loss": 0.1947,
+      "step": 8406
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0327212806026367e-05,
+      "loss": 0.1521,
+      "step": 8407
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0324858757062148e-05,
+      "loss": 0.0927,
+      "step": 8408
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0322504708097932e-05,
+      "loss": 0.134,
+      "step": 8409
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0320150659133712e-05,
+      "loss": 0.1424,
+      "step": 8410
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.031779661016949e-05,
+      "loss": 0.0999,
+      "step": 8411
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0315442561205277e-05,
+      "loss": 0.1652,
+      "step": 8412
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0313088512241057e-05,
+      "loss": 0.2822,
+      "step": 8413
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0310734463276834e-05,
+      "loss": 0.0639,
+      "step": 8414
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0308380414312622e-05,
+      "loss": 0.11,
+      "step": 8415
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.03060263653484e-05,
+      "loss": 0.1407,
+      "step": 8416
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.030367231638418e-05,
+      "loss": 0.0793,
+      "step": 8417
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0301318267419963e-05,
+      "loss": 0.1893,
+      "step": 8418
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0298964218455744e-05,
+      "loss": 0.2012,
+      "step": 8419
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0296610169491528e-05,
+      "loss": 0.1755,
+      "step": 8420
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.029425612052731e-05,
+      "loss": 0.1215,
+      "step": 8421
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.029190207156309e-05,
+      "loss": 0.2062,
+      "step": 8422
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0289548022598873e-05,
+      "loss": 0.1313,
+      "step": 8423
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0287193973634653e-05,
+      "loss": 0.0466,
+      "step": 8424
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0284839924670434e-05,
+      "loss": 0.0858,
+      "step": 8425
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0282485875706218e-05,
+      "loss": 0.0848,
+      "step": 8426
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0280131826742e-05,
+      "loss": 0.067,
+      "step": 8427
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0277777777777776e-05,
+      "loss": 0.2886,
+      "step": 8428
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0275423728813563e-05,
+      "loss": 0.1309,
+      "step": 8429
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0273069679849343e-05,
+      "loss": 0.2039,
+      "step": 8430
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.027071563088512e-05,
+      "loss": 0.1765,
+      "step": 8431
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0268361581920908e-05,
+      "loss": 0.0662,
+      "step": 8432
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0266007532956685e-05,
+      "loss": 0.2578,
+      "step": 8433
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0263653483992472e-05,
+      "loss": 0.1832,
+      "step": 8434
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.026129943502825e-05,
+      "loss": 0.0601,
+      "step": 8435
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.025894538606403e-05,
+      "loss": 0.2289,
+      "step": 8436
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0256591337099814e-05,
+      "loss": 0.1915,
+      "step": 8437
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0254237288135594e-05,
+      "loss": 0.2242,
+      "step": 8438
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0251883239171375e-05,
+      "loss": 0.212,
+      "step": 8439
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.024952919020716e-05,
+      "loss": 0.0735,
+      "step": 8440
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.024717514124294e-05,
+      "loss": 0.1321,
+      "step": 8441
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.024482109227872e-05,
+      "loss": 0.0504,
+      "step": 8442
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0242467043314504e-05,
+      "loss": 0.1578,
+      "step": 8443
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0240112994350284e-05,
+      "loss": 0.243,
+      "step": 8444
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.023775894538606e-05,
+      "loss": 0.085,
+      "step": 8445
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.023540489642185e-05,
+      "loss": 0.0813,
+      "step": 8446
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.023305084745763e-05,
+      "loss": 0.1924,
+      "step": 8447
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0230696798493407e-05,
+      "loss": 0.1499,
+      "step": 8448
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0228342749529194e-05,
+      "loss": 0.2123,
+      "step": 8449
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.022598870056497e-05,
+      "loss": 0.0282,
+      "step": 8450
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.022363465160076e-05,
+      "loss": 0.1924,
+      "step": 8451
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0221280602636535e-05,
+      "loss": 0.1416,
+      "step": 8452
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0218926553672316e-05,
+      "loss": 0.119,
+      "step": 8453
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0216572504708103e-05,
+      "loss": 0.1332,
+      "step": 8454
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.021421845574388e-05,
+      "loss": 0.1465,
+      "step": 8455
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.021186440677966e-05,
+      "loss": 0.2246,
+      "step": 8456
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0209510357815445e-05,
+      "loss": 0.19,
+      "step": 8457
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0207156308851225e-05,
+      "loss": 0.0506,
+      "step": 8458
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0204802259887006e-05,
+      "loss": 0.212,
+      "step": 8459
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.020244821092279e-05,
+      "loss": 0.1073,
+      "step": 8460
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.020009416195857e-05,
+      "loss": 0.1819,
+      "step": 8461
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.019774011299435e-05,
+      "loss": 0.1315,
+      "step": 8462
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0195386064030135e-05,
+      "loss": 0.1793,
+      "step": 8463
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0193032015065915e-05,
+      "loss": 0.1662,
+      "step": 8464
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0190677966101693e-05,
+      "loss": 0.0905,
+      "step": 8465
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.018832391713748e-05,
+      "loss": 0.067,
+      "step": 8466
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0185969868173257e-05,
+      "loss": 0.1639,
+      "step": 8467
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0183615819209044e-05,
+      "loss": 0.1987,
+      "step": 8468
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.018126177024482e-05,
+      "loss": 0.1946,
+      "step": 8469
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0178907721280602e-05,
+      "loss": 0.0846,
+      "step": 8470
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.017655367231639e-05,
+      "loss": 0.2406,
+      "step": 8471
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0174199623352167e-05,
+      "loss": 0.2638,
+      "step": 8472
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0171845574387947e-05,
+      "loss": 0.0955,
+      "step": 8473
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.016949152542373e-05,
+      "loss": 0.0956,
+      "step": 8474
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.016713747645951e-05,
+      "loss": 0.1178,
+      "step": 8475
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0164783427495292e-05,
+      "loss": 0.1073,
+      "step": 8476
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0162429378531076e-05,
+      "loss": 0.0985,
+      "step": 8477
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0160075329566857e-05,
+      "loss": 0.1716,
+      "step": 8478
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0157721280602637e-05,
+      "loss": 0.1287,
+      "step": 8479
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.015536723163842e-05,
+      "loss": 0.1144,
+      "step": 8480
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.01530131826742e-05,
+      "loss": 0.1117,
+      "step": 8481
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0150659133709985e-05,
+      "loss": 0.2271,
+      "step": 8482
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0148305084745766e-05,
+      "loss": 0.0971,
+      "step": 8483
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0145951035781543e-05,
+      "loss": 0.1256,
+      "step": 8484
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.014359698681733e-05,
+      "loss": 0.2122,
+      "step": 8485
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0141242937853108e-05,
+      "loss": 0.1119,
+      "step": 8486
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0138888888888888e-05,
+      "loss": 0.0868,
+      "step": 8487
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0136534839924675e-05,
+      "loss": 0.2098,
+      "step": 8488
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0134180790960453e-05,
+      "loss": 0.1234,
+      "step": 8489
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0131826741996233e-05,
+      "loss": 0.0686,
+      "step": 8490
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0129472693032017e-05,
+      "loss": 0.0684,
+      "step": 8491
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0127118644067798e-05,
+      "loss": 0.1562,
+      "step": 8492
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0124764595103578e-05,
+      "loss": 0.0819,
+      "step": 8493
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0122410546139362e-05,
+      "loss": 0.1187,
+      "step": 8494
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0120056497175143e-05,
+      "loss": 0.1182,
+      "step": 8495
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0117702448210923e-05,
+      "loss": 0.0379,
+      "step": 8496
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0115348399246707e-05,
+      "loss": 0.1361,
+      "step": 8497
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0112994350282488e-05,
+      "loss": 0.0876,
+      "step": 8498
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.011064030131827e-05,
+      "loss": 0.1144,
+      "step": 8499
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0108286252354052e-05,
+      "loss": 0.1565,
+      "step": 8500
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.010593220338983e-05,
+      "loss": 0.0971,
+      "step": 8501
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0103578154425617e-05,
+      "loss": 0.2312,
+      "step": 8502
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0101224105461394e-05,
+      "loss": 0.2127,
+      "step": 8503
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0098870056497174e-05,
+      "loss": 0.0696,
+      "step": 8504
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.009651600753296e-05,
+      "loss": 0.0553,
+      "step": 8505
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.009416195856874e-05,
+      "loss": 0.1047,
+      "step": 8506
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.009180790960452e-05,
+      "loss": 0.1628,
+      "step": 8507
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0089453860640303e-05,
+      "loss": 0.2126,
+      "step": 8508
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0087099811676084e-05,
+      "loss": 0.2575,
+      "step": 8509
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0084745762711864e-05,
+      "loss": 0.1546,
+      "step": 8510
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0082391713747648e-05,
+      "loss": 0.1938,
+      "step": 8511
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.008003766478343e-05,
+      "loss": 0.1291,
+      "step": 8512
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0077683615819213e-05,
+      "loss": 0.292,
+      "step": 8513
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0075329566854993e-05,
+      "loss": 0.0751,
+      "step": 8514
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0072975517890774e-05,
+      "loss": 0.1513,
+      "step": 8515
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0070621468926558e-05,
+      "loss": 0.0358,
+      "step": 8516
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.8926966787067219,
+      "eval_loss": 0.3247721791267395,
+      "eval_runtime": 22.3322,
+      "eval_samples_per_second": 1524.836,
+      "eval_steps_per_second": 23.867,
+      "step": 8516
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0068267419962338e-05,
+      "loss": 0.0634,
+      "step": 8517
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0065913370998115e-05,
+      "loss": 0.2071,
+      "step": 8518
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0063559322033903e-05,
+      "loss": 0.0486,
+      "step": 8519
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.006120527306968e-05,
+      "loss": 0.0249,
+      "step": 8520
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.005885122410546e-05,
+      "loss": 0.082,
+      "step": 8521
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0056497175141248e-05,
+      "loss": 0.0443,
+      "step": 8522
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0054143126177025e-05,
+      "loss": 0.0455,
+      "step": 8523
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0051789077212805e-05,
+      "loss": 0.0709,
+      "step": 8524
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.004943502824859e-05,
+      "loss": 0.0997,
+      "step": 8525
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.004708097928437e-05,
+      "loss": 0.1072,
+      "step": 8526
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.004472693032015e-05,
+      "loss": 0.1942,
+      "step": 8527
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.0042372881355934e-05,
+      "loss": 0.057,
+      "step": 8528
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.0040018832391715e-05,
+      "loss": 0.2017,
+      "step": 8529
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.00376647834275e-05,
+      "loss": 0.0528,
+      "step": 8530
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.003531073446328e-05,
+      "loss": 0.0201,
+      "step": 8531
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.003295668549906e-05,
+      "loss": 0.0521,
+      "step": 8532
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.0030602636534844e-05,
+      "loss": 0.019,
+      "step": 8533
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.0028248587570624e-05,
+      "loss": 0.1112,
+      "step": 8534
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.00258945386064e-05,
+      "loss": 0.0888,
+      "step": 8535
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.002354048964219e-05,
+      "loss": 0.0747,
+      "step": 8536
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.002118644067797e-05,
+      "loss": 0.0337,
+      "step": 8537
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.0018832391713746e-05,
+      "loss": 0.0449,
+      "step": 8538
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.0016478342749534e-05,
+      "loss": 0.017,
+      "step": 8539
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.001412429378531e-05,
+      "loss": 0.1217,
+      "step": 8540
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.001177024482109e-05,
+      "loss": 0.1419,
+      "step": 8541
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.0009416195856875e-05,
+      "loss": 0.079,
+      "step": 8542
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.0007062146892656e-05,
+      "loss": 0.0734,
+      "step": 8543
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.0004708097928436e-05,
+      "loss": 0.0547,
+      "step": 8544
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.000235404896422e-05,
+      "loss": 0.1253,
+      "step": 8545
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3e-05,
+      "loss": 0.065,
+      "step": 8546
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9997645951035785e-05,
+      "loss": 0.0814,
+      "step": 8547
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9995291902071565e-05,
+      "loss": 0.3794,
+      "step": 8548
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9992937853107346e-05,
+      "loss": 0.112,
+      "step": 8549
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.999058380414313e-05,
+      "loss": 0.1036,
+      "step": 8550
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.998822975517891e-05,
+      "loss": 0.0452,
+      "step": 8551
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9985875706214687e-05,
+      "loss": 0.1697,
+      "step": 8552
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9983521657250475e-05,
+      "loss": 0.0769,
+      "step": 8553
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9981167608286255e-05,
+      "loss": 0.0864,
+      "step": 8554
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9978813559322032e-05,
+      "loss": 0.0885,
+      "step": 8555
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.997645951035782e-05,
+      "loss": 0.1011,
+      "step": 8556
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9974105461393597e-05,
+      "loss": 0.054,
+      "step": 8557
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9971751412429377e-05,
+      "loss": 0.1309,
+      "step": 8558
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.996939736346516e-05,
+      "loss": 0.0404,
+      "step": 8559
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9967043314500942e-05,
+      "loss": 0.0335,
+      "step": 8560
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9964689265536726e-05,
+      "loss": 0.0614,
+      "step": 8561
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9962335216572506e-05,
+      "loss": 0.0502,
+      "step": 8562
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9959981167608287e-05,
+      "loss": 0.1289,
+      "step": 8563
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.995762711864407e-05,
+      "loss": 0.2044,
+      "step": 8564
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.995527306967985e-05,
+      "loss": 0.0663,
+      "step": 8565
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9952919020715632e-05,
+      "loss": 0.0317,
+      "step": 8566
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9950564971751416e-05,
+      "loss": 0.0943,
+      "step": 8567
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9948210922787196e-05,
+      "loss": 0.1003,
+      "step": 8568
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9945856873822973e-05,
+      "loss": 0.1286,
+      "step": 8569
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.994350282485876e-05,
+      "loss": 0.0136,
+      "step": 8570
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.994114877589454e-05,
+      "loss": 0.0645,
+      "step": 8571
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.993879472693032e-05,
+      "loss": 0.0481,
+      "step": 8572
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9936440677966106e-05,
+      "loss": 0.0267,
+      "step": 8573
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9934086629001883e-05,
+      "loss": 0.1774,
+      "step": 8574
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9931732580037663e-05,
+      "loss": 0.1377,
+      "step": 8575
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9929378531073447e-05,
+      "loss": 0.1238,
+      "step": 8576
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9927024482109228e-05,
+      "loss": 0.1158,
+      "step": 8577
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9924670433145012e-05,
+      "loss": 0.1805,
+      "step": 8578
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9922316384180792e-05,
+      "loss": 0.083,
+      "step": 8579
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9919962335216573e-05,
+      "loss": 0.0498,
+      "step": 8580
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9917608286252357e-05,
+      "loss": 0.072,
+      "step": 8581
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9915254237288137e-05,
+      "loss": 0.2418,
+      "step": 8582
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9912900188323918e-05,
+      "loss": 0.0803,
+      "step": 8583
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9910546139359702e-05,
+      "loss": 0.0883,
+      "step": 8584
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9908192090395482e-05,
+      "loss": 0.0573,
+      "step": 8585
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.990583804143126e-05,
+      "loss": 0.2116,
+      "step": 8586
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9903483992467047e-05,
+      "loss": 0.132,
+      "step": 8587
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9901129943502827e-05,
+      "loss": 0.0535,
+      "step": 8588
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9898775894538604e-05,
+      "loss": 0.1452,
+      "step": 8589
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9896421845574392e-05,
+      "loss": 0.0855,
+      "step": 8590
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.989406779661017e-05,
+      "loss": 0.0886,
+      "step": 8591
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9891713747645956e-05,
+      "loss": 0.0914,
+      "step": 8592
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9889359698681733e-05,
+      "loss": 0.0787,
+      "step": 8593
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9887005649717514e-05,
+      "loss": 0.029,
+      "step": 8594
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9884651600753298e-05,
+      "loss": 0.0532,
+      "step": 8595
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.988229755178908e-05,
+      "loss": 0.0509,
+      "step": 8596
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.987994350282486e-05,
+      "loss": 0.1548,
+      "step": 8597
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9877589453860643e-05,
+      "loss": 0.0863,
+      "step": 8598
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9875235404896423e-05,
+      "loss": 0.2276,
+      "step": 8599
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9872881355932204e-05,
+      "loss": 0.0166,
+      "step": 8600
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9870527306967988e-05,
+      "loss": 0.0388,
+      "step": 8601
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.986817325800377e-05,
+      "loss": 0.0646,
+      "step": 8602
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9865819209039546e-05,
+      "loss": 0.0889,
+      "step": 8603
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9863465160075333e-05,
+      "loss": 0.0636,
+      "step": 8604
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9861111111111113e-05,
+      "loss": 0.0249,
+      "step": 8605
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.985875706214689e-05,
+      "loss": 0.0408,
+      "step": 8606
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9856403013182678e-05,
+      "loss": 0.0729,
+      "step": 8607
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9854048964218455e-05,
+      "loss": 0.1156,
+      "step": 8608
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9851694915254242e-05,
+      "loss": 0.1175,
+      "step": 8609
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.984934086629002e-05,
+      "loss": 0.0117,
+      "step": 8610
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.98469868173258e-05,
+      "loss": 0.1403,
+      "step": 8611
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9844632768361587e-05,
+      "loss": 0.0447,
+      "step": 8612
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9842278719397364e-05,
+      "loss": 0.1562,
+      "step": 8613
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9839924670433145e-05,
+      "loss": 0.0376,
+      "step": 8614
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.983757062146893e-05,
+      "loss": 0.0515,
+      "step": 8615
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.983521657250471e-05,
+      "loss": 0.0242,
+      "step": 8616
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.983286252354049e-05,
+      "loss": 0.0866,
+      "step": 8617
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9830508474576274e-05,
+      "loss": 0.1524,
+      "step": 8618
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9828154425612054e-05,
+      "loss": 0.0525,
+      "step": 8619
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9825800376647835e-05,
+      "loss": 0.0386,
+      "step": 8620
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.982344632768362e-05,
+      "loss": 0.1846,
+      "step": 8621
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.98210922787194e-05,
+      "loss": 0.0905,
+      "step": 8622
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9818738229755177e-05,
+      "loss": 0.0982,
+      "step": 8623
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9816384180790964e-05,
+      "loss": 0.2067,
+      "step": 8624
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.981403013182674e-05,
+      "loss": 0.1465,
+      "step": 8625
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.981167608286253e-05,
+      "loss": 0.0335,
+      "step": 8626
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9809322033898306e-05,
+      "loss": 0.0274,
+      "step": 8627
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9806967984934086e-05,
+      "loss": 0.3188,
+      "step": 8628
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9804613935969873e-05,
+      "loss": 0.0565,
+      "step": 8629
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.980225988700565e-05,
+      "loss": 0.1301,
+      "step": 8630
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.979990583804143e-05,
+      "loss": 0.0464,
+      "step": 8631
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9797551789077215e-05,
+      "loss": 0.0714,
+      "step": 8632
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9795197740112996e-05,
+      "loss": 0.1273,
+      "step": 8633
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9792843691148776e-05,
+      "loss": 0.2141,
+      "step": 8634
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.979048964218456e-05,
+      "loss": 0.0379,
+      "step": 8635
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.978813559322034e-05,
+      "loss": 0.0651,
+      "step": 8636
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.978578154425612e-05,
+      "loss": 0.0419,
+      "step": 8637
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9783427495291905e-05,
+      "loss": 0.0836,
+      "step": 8638
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9781073446327686e-05,
+      "loss": 0.0978,
+      "step": 8639
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.977871939736347e-05,
+      "loss": 0.1786,
+      "step": 8640
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.977636534839925e-05,
+      "loss": 0.1485,
+      "step": 8641
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9774011299435027e-05,
+      "loss": 0.0491,
+      "step": 8642
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9771657250470814e-05,
+      "loss": 0.0941,
+      "step": 8643
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.976930320150659e-05,
+      "loss": 0.0713,
+      "step": 8644
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9766949152542372e-05,
+      "loss": 0.2227,
+      "step": 8645
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.976459510357816e-05,
+      "loss": 0.1413,
+      "step": 8646
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9762241054613937e-05,
+      "loss": 0.066,
+      "step": 8647
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9759887005649717e-05,
+      "loss": 0.1347,
+      "step": 8648
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.97575329566855e-05,
+      "loss": 0.1324,
+      "step": 8649
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.975517890772128e-05,
+      "loss": 0.1908,
+      "step": 8650
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9752824858757062e-05,
+      "loss": 0.0597,
+      "step": 8651
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9750470809792846e-05,
+      "loss": 0.119,
+      "step": 8652
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9748116760828627e-05,
+      "loss": 0.1525,
+      "step": 8653
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9745762711864407e-05,
+      "loss": 0.1266,
+      "step": 8654
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.974340866290019e-05,
+      "loss": 0.0824,
+      "step": 8655
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.974105461393597e-05,
+      "loss": 0.0704,
+      "step": 8656
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9738700564971755e-05,
+      "loss": 0.0415,
+      "step": 8657
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9736346516007536e-05,
+      "loss": 0.0716,
+      "step": 8658
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9733992467043313e-05,
+      "loss": 0.0619,
+      "step": 8659
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.97316384180791e-05,
+      "loss": 0.2959,
+      "step": 8660
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9729284369114878e-05,
+      "loss": 0.0224,
+      "step": 8661
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9726930320150658e-05,
+      "loss": 0.2779,
+      "step": 8662
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9724576271186445e-05,
+      "loss": 0.1095,
+      "step": 8663
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9722222222222223e-05,
+      "loss": 0.0968,
+      "step": 8664
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9719868173258003e-05,
+      "loss": 0.1045,
+      "step": 8665
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9717514124293787e-05,
+      "loss": 0.083,
+      "step": 8666
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9715160075329568e-05,
+      "loss": 0.0787,
+      "step": 8667
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9712806026365348e-05,
+      "loss": 0.0544,
+      "step": 8668
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9710451977401132e-05,
+      "loss": 0.0772,
+      "step": 8669
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9708097928436913e-05,
+      "loss": 0.0354,
+      "step": 8670
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9705743879472697e-05,
+      "loss": 0.0206,
+      "step": 8671
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9703389830508477e-05,
+      "loss": 0.1106,
+      "step": 8672
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9701035781544258e-05,
+      "loss": 0.1318,
+      "step": 8673
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.969868173258004e-05,
+      "loss": 0.1318,
+      "step": 8674
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9696327683615822e-05,
+      "loss": 0.2421,
+      "step": 8675
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.96939736346516e-05,
+      "loss": 0.0719,
+      "step": 8676
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9691619585687387e-05,
+      "loss": 0.1387,
+      "step": 8677
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9689265536723164e-05,
+      "loss": 0.1926,
+      "step": 8678
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9686911487758944e-05,
+      "loss": 0.0252,
+      "step": 8679
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.968455743879473e-05,
+      "loss": 0.0543,
+      "step": 8680
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.968220338983051e-05,
+      "loss": 0.0489,
+      "step": 8681
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.967984934086629e-05,
+      "loss": 0.0648,
+      "step": 8682
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9677495291902073e-05,
+      "loss": 0.1365,
+      "step": 8683
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9675141242937854e-05,
+      "loss": 0.1176,
+      "step": 8684
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9672787193973634e-05,
+      "loss": 0.1435,
+      "step": 8685
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9670433145009418e-05,
+      "loss": 0.1502,
+      "step": 8686
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.96680790960452e-05,
+      "loss": 0.1534,
+      "step": 8687
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9665725047080983e-05,
+      "loss": 0.0932,
+      "step": 8688
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9663370998116763e-05,
+      "loss": 0.1264,
+      "step": 8689
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9661016949152544e-05,
+      "loss": 0.2308,
+      "step": 8690
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9658662900188328e-05,
+      "loss": 0.1161,
+      "step": 8691
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9656308851224108e-05,
+      "loss": 0.1346,
+      "step": 8692
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9653954802259885e-05,
+      "loss": 0.1168,
+      "step": 8693
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9651600753295673e-05,
+      "loss": 0.0538,
+      "step": 8694
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9649246704331453e-05,
+      "loss": 0.117,
+      "step": 8695
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.964689265536723e-05,
+      "loss": 0.1956,
+      "step": 8696
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9644538606403018e-05,
+      "loss": 0.2204,
+      "step": 8697
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9642184557438795e-05,
+      "loss": 0.1086,
+      "step": 8698
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9639830508474575e-05,
+      "loss": 0.0371,
+      "step": 8699
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.963747645951036e-05,
+      "loss": 0.0701,
+      "step": 8700
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.963512241054614e-05,
+      "loss": 0.1004,
+      "step": 8701
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.963276836158192e-05,
+      "loss": 0.0317,
+      "step": 8702
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9630414312617704e-05,
+      "loss": 0.1161,
+      "step": 8703
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9628060263653485e-05,
+      "loss": 0.0527,
+      "step": 8704
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.962570621468927e-05,
+      "loss": 0.1598,
+      "step": 8705
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.962335216572505e-05,
+      "loss": 0.0227,
+      "step": 8706
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.962099811676083e-05,
+      "loss": 0.0792,
+      "step": 8707
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9618644067796614e-05,
+      "loss": 0.1895,
+      "step": 8708
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9616290018832394e-05,
+      "loss": 0.07,
+      "step": 8709
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.961393596986817e-05,
+      "loss": 0.027,
+      "step": 8710
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.961158192090396e-05,
+      "loss": 0.0352,
+      "step": 8711
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.960922787193974e-05,
+      "loss": 0.0978,
+      "step": 8712
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9606873822975516e-05,
+      "loss": 0.1551,
+      "step": 8713
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9604519774011304e-05,
+      "loss": 0.1374,
+      "step": 8714
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.960216572504708e-05,
+      "loss": 0.1507,
+      "step": 8715
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.959981167608286e-05,
+      "loss": 0.0994,
+      "step": 8716
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9597457627118645e-05,
+      "loss": 0.1631,
+      "step": 8717
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9595103578154426e-05,
+      "loss": 0.0396,
+      "step": 8718
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.959274952919021e-05,
+      "loss": 0.1043,
+      "step": 8719
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.959039548022599e-05,
+      "loss": 0.1633,
+      "step": 8720
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.958804143126177e-05,
+      "loss": 0.1061,
+      "step": 8721
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9585687382297555e-05,
+      "loss": 0.0263,
+      "step": 8722
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9583333333333335e-05,
+      "loss": 0.0191,
+      "step": 8723
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9580979284369116e-05,
+      "loss": 0.0663,
+      "step": 8724
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.95786252354049e-05,
+      "loss": 0.1333,
+      "step": 8725
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.957627118644068e-05,
+      "loss": 0.1165,
+      "step": 8726
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9573917137476457e-05,
+      "loss": 0.1447,
+      "step": 8727
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9571563088512245e-05,
+      "loss": 0.0561,
+      "step": 8728
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9569209039548025e-05,
+      "loss": 0.0788,
+      "step": 8729
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9566854990583802e-05,
+      "loss": 0.1295,
+      "step": 8730
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.956450094161959e-05,
+      "loss": 0.0515,
+      "step": 8731
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9562146892655367e-05,
+      "loss": 0.0612,
+      "step": 8732
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9559792843691147e-05,
+      "loss": 0.1022,
+      "step": 8733
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.955743879472693e-05,
+      "loss": 0.0264,
+      "step": 8734
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9555084745762712e-05,
+      "loss": 0.0145,
+      "step": 8735
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9552730696798496e-05,
+      "loss": 0.0948,
+      "step": 8736
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9550376647834276e-05,
+      "loss": 0.1469,
+      "step": 8737
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9548022598870057e-05,
+      "loss": 0.2011,
+      "step": 8738
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.954566854990584e-05,
+      "loss": 0.0958,
+      "step": 8739
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.954331450094162e-05,
+      "loss": 0.023,
+      "step": 8740
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9540960451977402e-05,
+      "loss": 0.0483,
+      "step": 8741
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9538606403013186e-05,
+      "loss": 0.0734,
+      "step": 8742
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9536252354048966e-05,
+      "loss": 0.0459,
+      "step": 8743
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9533898305084743e-05,
+      "loss": 0.1634,
+      "step": 8744
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.953154425612053e-05,
+      "loss": 0.1892,
+      "step": 8745
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.952919020715631e-05,
+      "loss": 0.0549,
+      "step": 8746
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.952683615819209e-05,
+      "loss": 0.0385,
+      "step": 8747
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9524482109227876e-05,
+      "loss": 0.0204,
+      "step": 8748
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9522128060263653e-05,
+      "loss": 0.0501,
+      "step": 8749
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.951977401129944e-05,
+      "loss": 0.1281,
+      "step": 8750
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9517419962335217e-05,
+      "loss": 0.1295,
+      "step": 8751
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9515065913370998e-05,
+      "loss": 0.0952,
+      "step": 8752
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9512711864406782e-05,
+      "loss": 0.0491,
+      "step": 8753
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9510357815442562e-05,
+      "loss": 0.0319,
+      "step": 8754
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9508003766478343e-05,
+      "loss": 0.1954,
+      "step": 8755
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9505649717514127e-05,
+      "loss": 0.1562,
+      "step": 8756
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9503295668549907e-05,
+      "loss": 0.0613,
+      "step": 8757
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9500941619585688e-05,
+      "loss": 0.1259,
+      "step": 8758
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9498587570621472e-05,
+      "loss": 0.2717,
+      "step": 8759
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9496233521657252e-05,
+      "loss": 0.1079,
+      "step": 8760
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.949387947269303e-05,
+      "loss": 0.0336,
+      "step": 8761
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9491525423728817e-05,
+      "loss": 0.1769,
+      "step": 8762
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9489171374764597e-05,
+      "loss": 0.0499,
+      "step": 8763
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9486817325800375e-05,
+      "loss": 0.0683,
+      "step": 8764
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9484463276836162e-05,
+      "loss": 0.0556,
+      "step": 8765
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.948210922787194e-05,
+      "loss": 0.1492,
+      "step": 8766
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9479755178907726e-05,
+      "loss": 0.0962,
+      "step": 8767
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9477401129943503e-05,
+      "loss": 0.0286,
+      "step": 8768
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9475047080979284e-05,
+      "loss": 0.0135,
+      "step": 8769
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.947269303201507e-05,
+      "loss": 0.1931,
+      "step": 8770
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.947033898305085e-05,
+      "loss": 0.0332,
+      "step": 8771
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.946798493408663e-05,
+      "loss": 0.0307,
+      "step": 8772
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9465630885122413e-05,
+      "loss": 0.127,
+      "step": 8773
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9463276836158193e-05,
+      "loss": 0.0279,
+      "step": 8774
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9460922787193974e-05,
+      "loss": 0.1046,
+      "step": 8775
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9458568738229758e-05,
+      "loss": 0.0789,
+      "step": 8776
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.945621468926554e-05,
+      "loss": 0.142,
+      "step": 8777
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.945386064030132e-05,
+      "loss": 0.0697,
+      "step": 8778
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9451506591337103e-05,
+      "loss": 0.1657,
+      "step": 8779
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9449152542372883e-05,
+      "loss": 0.066,
+      "step": 8780
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.944679849340866e-05,
+      "loss": 0.0597,
+      "step": 8781
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9444444444444448e-05,
+      "loss": 0.076,
+      "step": 8782
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9442090395480225e-05,
+      "loss": 0.1334,
+      "step": 8783
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9439736346516012e-05,
+      "loss": 0.0319,
+      "step": 8784
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.943738229755179e-05,
+      "loss": 0.0521,
+      "step": 8785
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.943502824858757e-05,
+      "loss": 0.0682,
+      "step": 8786
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9432674199623357e-05,
+      "loss": 0.0452,
+      "step": 8787
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9430320150659134e-05,
+      "loss": 0.0751,
+      "step": 8788
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9427966101694915e-05,
+      "loss": 0.0268,
+      "step": 8789
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.94256120527307e-05,
+      "loss": 0.0854,
+      "step": 8790
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.942325800376648e-05,
+      "loss": 0.1361,
+      "step": 8791
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.942090395480226e-05,
+      "loss": 0.1136,
+      "step": 8792
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9418549905838044e-05,
+      "loss": 0.1534,
+      "step": 8793
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9416195856873824e-05,
+      "loss": 0.1363,
+      "step": 8794
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9413841807909605e-05,
+      "loss": 0.1035,
+      "step": 8795
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.941148775894539e-05,
+      "loss": 0.1171,
+      "step": 8796
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.940913370998117e-05,
+      "loss": 0.079,
+      "step": 8797
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9406779661016953e-05,
+      "loss": 0.143,
+      "step": 8798
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9404425612052734e-05,
+      "loss": 0.1022,
+      "step": 8799
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.940207156308851e-05,
+      "loss": 0.0404,
+      "step": 8800
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.93997175141243e-05,
+      "loss": 0.0908,
+      "step": 8801
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9397363465160076e-05,
+      "loss": 0.0328,
+      "step": 8802
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9395009416195856e-05,
+      "loss": 0.0605,
+      "step": 8803
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9392655367231643e-05,
+      "loss": 0.0287,
+      "step": 8804
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.939030131826742e-05,
+      "loss": 0.1417,
+      "step": 8805
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.93879472693032e-05,
+      "loss": 0.0254,
+      "step": 8806
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9385593220338985e-05,
+      "loss": 0.0962,
+      "step": 8807
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9383239171374766e-05,
+      "loss": 0.1798,
+      "step": 8808
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9380885122410546e-05,
+      "loss": 0.0458,
+      "step": 8809
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.937853107344633e-05,
+      "loss": 0.148,
+      "step": 8810
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.937617702448211e-05,
+      "loss": 0.1405,
+      "step": 8811
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.937382297551789e-05,
+      "loss": 0.0812,
+      "step": 8812
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9371468926553675e-05,
+      "loss": 0.1373,
+      "step": 8813
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9369114877589456e-05,
+      "loss": 0.0576,
+      "step": 8814
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.936676082862524e-05,
+      "loss": 0.1887,
+      "step": 8815
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.936440677966102e-05,
+      "loss": 0.0743,
+      "step": 8816
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9362052730696797e-05,
+      "loss": 0.127,
+      "step": 8817
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9359698681732584e-05,
+      "loss": 0.0717,
+      "step": 8818
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.935734463276836e-05,
+      "loss": 0.1115,
+      "step": 8819
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9354990583804142e-05,
+      "loss": 0.0674,
+      "step": 8820
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.935263653483993e-05,
+      "loss": 0.1244,
+      "step": 8821
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9350282485875707e-05,
+      "loss": 0.1663,
+      "step": 8822
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9347928436911487e-05,
+      "loss": 0.056,
+      "step": 8823
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.934557438794727e-05,
+      "loss": 0.006,
+      "step": 8824
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.934322033898305e-05,
+      "loss": 0.0132,
+      "step": 8825
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9340866290018832e-05,
+      "loss": 0.1128,
+      "step": 8826
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9338512241054616e-05,
+      "loss": 0.0231,
+      "step": 8827
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9336158192090397e-05,
+      "loss": 0.2995,
+      "step": 8828
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.933380414312618e-05,
+      "loss": 0.051,
+      "step": 8829
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.933145009416196e-05,
+      "loss": 0.2008,
+      "step": 8830
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.932909604519774e-05,
+      "loss": 0.1113,
+      "step": 8831
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9326741996233526e-05,
+      "loss": 0.1044,
+      "step": 8832
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9324387947269306e-05,
+      "loss": 0.0732,
+      "step": 8833
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9322033898305083e-05,
+      "loss": 0.1571,
+      "step": 8834
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.931967984934087e-05,
+      "loss": 0.1676,
+      "step": 8835
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9317325800376648e-05,
+      "loss": 0.1725,
+      "step": 8836
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9314971751412428e-05,
+      "loss": 0.1367,
+      "step": 8837
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9312617702448216e-05,
+      "loss": 0.0377,
+      "step": 8838
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9310263653483993e-05,
+      "loss": 0.0501,
+      "step": 8839
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9307909604519773e-05,
+      "loss": 0.044,
+      "step": 8840
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9305555555555557e-05,
+      "loss": 0.0465,
+      "step": 8841
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9303201506591338e-05,
+      "loss": 0.1251,
+      "step": 8842
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9300847457627118e-05,
+      "loss": 0.1855,
+      "step": 8843
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9298493408662902e-05,
+      "loss": 0.1995,
+      "step": 8844
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9296139359698683e-05,
+      "loss": 0.0807,
+      "step": 8845
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9293785310734467e-05,
+      "loss": 0.0573,
+      "step": 8846
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9291431261770247e-05,
+      "loss": 0.0301,
+      "step": 8847
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9289077212806028e-05,
+      "loss": 0.0563,
+      "step": 8848
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.928672316384181e-05,
+      "loss": 0.023,
+      "step": 8849
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9284369114877592e-05,
+      "loss": 0.0828,
+      "step": 8850
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.928201506591337e-05,
+      "loss": 0.0716,
+      "step": 8851
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9279661016949157e-05,
+      "loss": 0.0269,
+      "step": 8852
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9277306967984937e-05,
+      "loss": 0.1356,
+      "step": 8853
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9274952919020714e-05,
+      "loss": 0.0698,
+      "step": 8854
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.92725988700565e-05,
+      "loss": 0.1402,
+      "step": 8855
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.927024482109228e-05,
+      "loss": 0.0111,
+      "step": 8856
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.926789077212806e-05,
+      "loss": 0.1386,
+      "step": 8857
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9265536723163843e-05,
+      "loss": 0.2795,
+      "step": 8858
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9263182674199624e-05,
+      "loss": 0.148,
+      "step": 8859
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9260828625235404e-05,
+      "loss": 0.1002,
+      "step": 8860
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9258474576271188e-05,
+      "loss": 0.097,
+      "step": 8861
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.925612052730697e-05,
+      "loss": 0.0255,
+      "step": 8862
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9253766478342753e-05,
+      "loss": 0.0529,
+      "step": 8863
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9251412429378533e-05,
+      "loss": 0.0508,
+      "step": 8864
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9249058380414314e-05,
+      "loss": 0.0689,
+      "step": 8865
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9246704331450098e-05,
+      "loss": 0.0888,
+      "step": 8866
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9244350282485878e-05,
+      "loss": 0.0239,
+      "step": 8867
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9241996233521655e-05,
+      "loss": 0.1525,
+      "step": 8868
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9239642184557443e-05,
+      "loss": 0.0404,
+      "step": 8869
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9237288135593223e-05,
+      "loss": 0.1752,
+      "step": 8870
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9234934086629e-05,
+      "loss": 0.127,
+      "step": 8871
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9232580037664788e-05,
+      "loss": 0.0409,
+      "step": 8872
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9230225988700565e-05,
+      "loss": 0.0661,
+      "step": 8873
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9227871939736345e-05,
+      "loss": 0.0183,
+      "step": 8874
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.922551789077213e-05,
+      "loss": 0.0293,
+      "step": 8875
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.922316384180791e-05,
+      "loss": 0.1002,
+      "step": 8876
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9220809792843694e-05,
+      "loss": 0.0461,
+      "step": 8877
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9218455743879474e-05,
+      "loss": 0.0711,
+      "step": 8878
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9216101694915255e-05,
+      "loss": 0.0379,
+      "step": 8879
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.921374764595104e-05,
+      "loss": 0.1535,
+      "step": 8880
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.921139359698682e-05,
+      "loss": 0.0707,
+      "step": 8881
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.92090395480226e-05,
+      "loss": 0.0757,
+      "step": 8882
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9206685499058384e-05,
+      "loss": 0.0495,
+      "step": 8883
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9204331450094164e-05,
+      "loss": 0.122,
+      "step": 8884
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.920197740112994e-05,
+      "loss": 0.0565,
+      "step": 8885
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.919962335216573e-05,
+      "loss": 0.0573,
+      "step": 8886
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.919726930320151e-05,
+      "loss": 0.0198,
+      "step": 8887
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9194915254237286e-05,
+      "loss": 0.0543,
+      "step": 8888
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9192561205273074e-05,
+      "loss": 0.0859,
+      "step": 8889
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.919020715630885e-05,
+      "loss": 0.0305,
+      "step": 8890
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.918785310734463e-05,
+      "loss": 0.1096,
+      "step": 8891
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9185499058380415e-05,
+      "loss": 0.0349,
+      "step": 8892
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9183145009416196e-05,
+      "loss": 0.0964,
+      "step": 8893
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.918079096045198e-05,
+      "loss": 0.051,
+      "step": 8894
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.917843691148776e-05,
+      "loss": 0.0329,
+      "step": 8895
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.917608286252354e-05,
+      "loss": 0.0376,
+      "step": 8896
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9173728813559325e-05,
+      "loss": 0.165,
+      "step": 8897
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9171374764595105e-05,
+      "loss": 0.1039,
+      "step": 8898
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9169020715630886e-05,
+      "loss": 0.0208,
+      "step": 8899
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.916666666666667e-05,
+      "loss": 0.0406,
+      "step": 8900
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.916431261770245e-05,
+      "loss": 0.1065,
+      "step": 8901
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9161958568738227e-05,
+      "loss": 0.0227,
+      "step": 8902
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9159604519774015e-05,
+      "loss": 0.1666,
+      "step": 8903
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9157250470809795e-05,
+      "loss": 0.0758,
+      "step": 8904
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9154896421845572e-05,
+      "loss": 0.1271,
+      "step": 8905
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.915254237288136e-05,
+      "loss": 0.044,
+      "step": 8906
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9150188323917137e-05,
+      "loss": 0.0198,
+      "step": 8907
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9147834274952924e-05,
+      "loss": 0.087,
+      "step": 8908
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.91454802259887e-05,
+      "loss": 0.0328,
+      "step": 8909
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9143126177024482e-05,
+      "loss": 0.1287,
+      "step": 8910
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9140772128060266e-05,
+      "loss": 0.0127,
+      "step": 8911
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9138418079096046e-05,
+      "loss": 0.0325,
+      "step": 8912
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9136064030131827e-05,
+      "loss": 0.125,
+      "step": 8913
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.913370998116761e-05,
+      "loss": 0.0467,
+      "step": 8914
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.913135593220339e-05,
+      "loss": 0.0785,
+      "step": 8915
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9129001883239172e-05,
+      "loss": 0.2713,
+      "step": 8916
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9126647834274956e-05,
+      "loss": 0.0466,
+      "step": 8917
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9124293785310736e-05,
+      "loss": 0.0816,
+      "step": 8918
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9121939736346514e-05,
+      "loss": 0.1739,
+      "step": 8919
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.91195856873823e-05,
+      "loss": 0.0265,
+      "step": 8920
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.911723163841808e-05,
+      "loss": 0.124,
+      "step": 8921
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.911487758945386e-05,
+      "loss": 0.1982,
+      "step": 8922
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9112523540489646e-05,
+      "loss": 0.1127,
+      "step": 8923
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9110169491525423e-05,
+      "loss": 0.0206,
+      "step": 8924
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.910781544256121e-05,
+      "loss": 0.0993,
+      "step": 8925
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9105461393596987e-05,
+      "loss": 0.106,
+      "step": 8926
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9103107344632768e-05,
+      "loss": 0.2007,
+      "step": 8927
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9100753295668552e-05,
+      "loss": 0.145,
+      "step": 8928
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9098399246704332e-05,
+      "loss": 0.0943,
+      "step": 8929
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9096045197740113e-05,
+      "loss": 0.0403,
+      "step": 8930
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9093691148775897e-05,
+      "loss": 0.0512,
+      "step": 8931
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9091337099811677e-05,
+      "loss": 0.0443,
+      "step": 8932
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9088983050847458e-05,
+      "loss": 0.0757,
+      "step": 8933
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9086629001883242e-05,
+      "loss": 0.1566,
+      "step": 8934
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9084274952919022e-05,
+      "loss": 0.0965,
+      "step": 8935
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9081920903954803e-05,
+      "loss": 0.129,
+      "step": 8936
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9079566854990587e-05,
+      "loss": 0.2269,
+      "step": 8937
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9077212806026367e-05,
+      "loss": 0.1197,
+      "step": 8938
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9074858757062145e-05,
+      "loss": 0.0915,
+      "step": 8939
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9072504708097932e-05,
+      "loss": 0.1919,
+      "step": 8940
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.907015065913371e-05,
+      "loss": 0.0934,
+      "step": 8941
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9067796610169496e-05,
+      "loss": 0.0796,
+      "step": 8942
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9065442561205273e-05,
+      "loss": 0.0717,
+      "step": 8943
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9063088512241054e-05,
+      "loss": 0.1326,
+      "step": 8944
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.906073446327684e-05,
+      "loss": 0.0222,
+      "step": 8945
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.905838041431262e-05,
+      "loss": 0.0096,
+      "step": 8946
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.90560263653484e-05,
+      "loss": 0.0416,
+      "step": 8947
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9053672316384183e-05,
+      "loss": 0.1167,
+      "step": 8948
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9051318267419963e-05,
+      "loss": 0.1803,
+      "step": 8949
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9048964218455744e-05,
+      "loss": 0.0513,
+      "step": 8950
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9046610169491528e-05,
+      "loss": 0.0699,
+      "step": 8951
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.904425612052731e-05,
+      "loss": 0.1219,
+      "step": 8952
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.904190207156309e-05,
+      "loss": 0.1147,
+      "step": 8953
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.9039548022598873e-05,
+      "loss": 0.081,
+      "step": 8954
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.9037193973634653e-05,
+      "loss": 0.1403,
+      "step": 8955
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.9034839924670437e-05,
+      "loss": 0.1858,
+      "step": 8956
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.9032485875706218e-05,
+      "loss": 0.1323,
+      "step": 8957
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.9030131826741995e-05,
+      "loss": 0.0512,
+      "step": 8958
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.9027777777777782e-05,
+      "loss": 0.1226,
+      "step": 8959
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.902542372881356e-05,
+      "loss": 0.1364,
+      "step": 8960
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.902306967984934e-05,
+      "loss": 0.0405,
+      "step": 8961
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.9020715630885127e-05,
+      "loss": 0.0461,
+      "step": 8962
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.9018361581920905e-05,
+      "loss": 0.1878,
+      "step": 8963
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.9016007532956685e-05,
+      "loss": 0.023,
+      "step": 8964
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.901365348399247e-05,
+      "loss": 0.09,
+      "step": 8965
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.901129943502825e-05,
+      "loss": 0.0751,
+      "step": 8966
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.900894538606403e-05,
+      "loss": 0.0487,
+      "step": 8967
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.9006591337099814e-05,
+      "loss": 0.0999,
+      "step": 8968
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.9004237288135595e-05,
+      "loss": 0.0826,
+      "step": 8969
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.9001883239171375e-05,
+      "loss": 0.0717,
+      "step": 8970
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.899952919020716e-05,
+      "loss": 0.1175,
+      "step": 8971
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.899717514124294e-05,
+      "loss": 0.1299,
+      "step": 8972
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8994821092278723e-05,
+      "loss": 0.0769,
+      "step": 8973
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8992467043314504e-05,
+      "loss": 0.1361,
+      "step": 8974
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.899011299435028e-05,
+      "loss": 0.1029,
+      "step": 8975
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.898775894538607e-05,
+      "loss": 0.0512,
+      "step": 8976
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8985404896421846e-05,
+      "loss": 0.1424,
+      "step": 8977
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8983050847457626e-05,
+      "loss": 0.0509,
+      "step": 8978
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8980696798493413e-05,
+      "loss": 0.0778,
+      "step": 8979
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.897834274952919e-05,
+      "loss": 0.0184,
+      "step": 8980
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.897598870056497e-05,
+      "loss": 0.0306,
+      "step": 8981
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8973634651600755e-05,
+      "loss": 0.1184,
+      "step": 8982
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8971280602636536e-05,
+      "loss": 0.02,
+      "step": 8983
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8968926553672316e-05,
+      "loss": 0.2259,
+      "step": 8984
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.89665725047081e-05,
+      "loss": 0.3256,
+      "step": 8985
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.896421845574388e-05,
+      "loss": 0.0518,
+      "step": 8986
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8961864406779665e-05,
+      "loss": 0.05,
+      "step": 8987
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8959510357815445e-05,
+      "loss": 0.1695,
+      "step": 8988
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8957156308851226e-05,
+      "loss": 0.0089,
+      "step": 8989
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.895480225988701e-05,
+      "loss": 0.1244,
+      "step": 8990
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.895244821092279e-05,
+      "loss": 0.1493,
+      "step": 8991
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8950094161958567e-05,
+      "loss": 0.0886,
+      "step": 8992
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8947740112994355e-05,
+      "loss": 0.1392,
+      "step": 8993
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.894538606403013e-05,
+      "loss": 0.1073,
+      "step": 8994
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8943032015065912e-05,
+      "loss": 0.1501,
+      "step": 8995
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.89406779661017e-05,
+      "loss": 0.0437,
+      "step": 8996
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8938323917137477e-05,
+      "loss": 0.0386,
+      "step": 8997
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8935969868173257e-05,
+      "loss": 0.0583,
+      "step": 8998
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.893361581920904e-05,
+      "loss": 0.1084,
+      "step": 8999
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.893126177024482e-05,
+      "loss": 0.2163,
+      "step": 9000
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8928907721280602e-05,
+      "loss": 0.07,
+      "step": 9001
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8926553672316386e-05,
+      "loss": 0.0716,
+      "step": 9002
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8924199623352167e-05,
+      "loss": 0.0717,
+      "step": 9003
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.892184557438795e-05,
+      "loss": 0.147,
+      "step": 9004
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.891949152542373e-05,
+      "loss": 0.221,
+      "step": 9005
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.891713747645951e-05,
+      "loss": 0.0645,
+      "step": 9006
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8914783427495296e-05,
+      "loss": 0.0847,
+      "step": 9007
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8912429378531076e-05,
+      "loss": 0.0586,
+      "step": 9008
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8910075329566853e-05,
+      "loss": 0.1272,
+      "step": 9009
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.890772128060264e-05,
+      "loss": 0.1302,
+      "step": 9010
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.890536723163842e-05,
+      "loss": 0.1946,
+      "step": 9011
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8903013182674198e-05,
+      "loss": 0.1303,
+      "step": 9012
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8900659133709986e-05,
+      "loss": 0.1252,
+      "step": 9013
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8898305084745763e-05,
+      "loss": 0.016,
+      "step": 9014
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8895951035781543e-05,
+      "loss": 0.0404,
+      "step": 9015
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8893596986817327e-05,
+      "loss": 0.0529,
+      "step": 9016
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8891242937853108e-05,
+      "loss": 0.1017,
+      "step": 9017
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8888888888888888e-05,
+      "loss": 0.0759,
+      "step": 9018
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8886534839924672e-05,
+      "loss": 0.1845,
+      "step": 9019
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8884180790960453e-05,
+      "loss": 0.0353,
+      "step": 9020
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8881826741996237e-05,
+      "loss": 0.0449,
+      "step": 9021
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8879472693032017e-05,
+      "loss": 0.0854,
+      "step": 9022
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8877118644067798e-05,
+      "loss": 0.1284,
+      "step": 9023
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.887476459510358e-05,
+      "loss": 0.1153,
+      "step": 9024
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8872410546139362e-05,
+      "loss": 0.0625,
+      "step": 9025
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.887005649717514e-05,
+      "loss": 0.0824,
+      "step": 9026
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8867702448210927e-05,
+      "loss": 0.1155,
+      "step": 9027
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8865348399246707e-05,
+      "loss": 0.0542,
+      "step": 9028
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8862994350282484e-05,
+      "loss": 0.1328,
+      "step": 9029
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.886064030131827e-05,
+      "loss": 0.2228,
+      "step": 9030
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.885828625235405e-05,
+      "loss": 0.0177,
+      "step": 9031
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.885593220338983e-05,
+      "loss": 0.127,
+      "step": 9032
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8853578154425613e-05,
+      "loss": 0.0249,
+      "step": 9033
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8851224105461394e-05,
+      "loss": 0.0255,
+      "step": 9034
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8848870056497178e-05,
+      "loss": 0.2226,
+      "step": 9035
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8846516007532958e-05,
+      "loss": 0.2327,
+      "step": 9036
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.884416195856874e-05,
+      "loss": 0.0598,
+      "step": 9037
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8841807909604523e-05,
+      "loss": 0.0474,
+      "step": 9038
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8839453860640303e-05,
+      "loss": 0.1016,
+      "step": 9039
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8837099811676084e-05,
+      "loss": 0.1852,
+      "step": 9040
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8834745762711868e-05,
+      "loss": 0.0976,
+      "step": 9041
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8832391713747648e-05,
+      "loss": 0.1059,
+      "step": 9042
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8830037664783425e-05,
+      "loss": 0.1136,
+      "step": 9043
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8827683615819213e-05,
+      "loss": 0.1403,
+      "step": 9044
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8825329566854993e-05,
+      "loss": 0.0964,
+      "step": 9045
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.882297551789077e-05,
+      "loss": 0.2125,
+      "step": 9046
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8820621468926558e-05,
+      "loss": 0.1516,
+      "step": 9047
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8818267419962335e-05,
+      "loss": 0.1014,
+      "step": 9048
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8815913370998115e-05,
+      "loss": 0.1231,
+      "step": 9049
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.88135593220339e-05,
+      "loss": 0.0786,
+      "step": 9050
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.881120527306968e-05,
+      "loss": 0.0522,
+      "step": 9051
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8808851224105464e-05,
+      "loss": 0.1173,
+      "step": 9052
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8806497175141244e-05,
+      "loss": 0.0655,
+      "step": 9053
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8804143126177025e-05,
+      "loss": 0.1832,
+      "step": 9054
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.880178907721281e-05,
+      "loss": 0.0258,
+      "step": 9055
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.879943502824859e-05,
+      "loss": 0.1454,
+      "step": 9056
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.879708097928437e-05,
+      "loss": 0.1565,
+      "step": 9057
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8794726930320154e-05,
+      "loss": 0.0833,
+      "step": 9058
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8792372881355934e-05,
+      "loss": 0.0619,
+      "step": 9059
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.879001883239171e-05,
+      "loss": 0.044,
+      "step": 9060
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.87876647834275e-05,
+      "loss": 0.2787,
+      "step": 9061
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.878531073446328e-05,
+      "loss": 0.0322,
+      "step": 9062
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8782956685499056e-05,
+      "loss": 0.0165,
+      "step": 9063
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8780602636534844e-05,
+      "loss": 0.0409,
+      "step": 9064
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.877824858757062e-05,
+      "loss": 0.0746,
+      "step": 9065
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8775894538606408e-05,
+      "loss": 0.0423,
+      "step": 9066
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8773540489642185e-05,
+      "loss": 0.196,
+      "step": 9067
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8771186440677966e-05,
+      "loss": 0.0403,
+      "step": 9068
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.876883239171375e-05,
+      "loss": 0.122,
+      "step": 9069
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.876647834274953e-05,
+      "loss": 0.1061,
+      "step": 9070
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.876412429378531e-05,
+      "loss": 0.0823,
+      "step": 9071
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8761770244821095e-05,
+      "loss": 0.1106,
+      "step": 9072
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8759416195856875e-05,
+      "loss": 0.0966,
+      "step": 9073
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8757062146892656e-05,
+      "loss": 0.102,
+      "step": 9074
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.875470809792844e-05,
+      "loss": 0.1276,
+      "step": 9075
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.875235404896422e-05,
+      "loss": 0.2188,
+      "step": 9076
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8749999999999997e-05,
+      "loss": 0.1175,
+      "step": 9077
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8747645951035785e-05,
+      "loss": 0.0418,
+      "step": 9078
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8745291902071565e-05,
+      "loss": 0.0481,
+      "step": 9079
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8742937853107342e-05,
+      "loss": 0.0218,
+      "step": 9080
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.874058380414313e-05,
+      "loss": 0.118,
+      "step": 9081
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8738229755178907e-05,
+      "loss": 0.0895,
+      "step": 9082
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8735875706214694e-05,
+      "loss": 0.1765,
+      "step": 9083
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.873352165725047e-05,
+      "loss": 0.0913,
+      "step": 9084
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8731167608286252e-05,
+      "loss": 0.024,
+      "step": 9085
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8728813559322036e-05,
+      "loss": 0.0566,
+      "step": 9086
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8726459510357816e-05,
+      "loss": 0.1033,
+      "step": 9087
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8724105461393597e-05,
+      "loss": 0.1674,
+      "step": 9088
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.872175141242938e-05,
+      "loss": 0.1792,
+      "step": 9089
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.871939736346516e-05,
+      "loss": 0.1865,
+      "step": 9090
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8717043314500942e-05,
+      "loss": 0.0693,
+      "step": 9091
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8714689265536726e-05,
+      "loss": 0.1559,
+      "step": 9092
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8712335216572506e-05,
+      "loss": 0.1135,
+      "step": 9093
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8709981167608287e-05,
+      "loss": 0.2215,
+      "step": 9094
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.870762711864407e-05,
+      "loss": 0.1243,
+      "step": 9095
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.870527306967985e-05,
+      "loss": 0.0804,
+      "step": 9096
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.870291902071563e-05,
+      "loss": 0.0451,
+      "step": 9097
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8700564971751416e-05,
+      "loss": 0.0529,
+      "step": 9098
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8698210922787193e-05,
+      "loss": 0.1574,
+      "step": 9099
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.869585687382298e-05,
+      "loss": 0.0481,
+      "step": 9100
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8693502824858757e-05,
+      "loss": 0.0297,
+      "step": 9101
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8691148775894538e-05,
+      "loss": 0.0742,
+      "step": 9102
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8688794726930325e-05,
+      "loss": 0.0757,
+      "step": 9103
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8686440677966102e-05,
+      "loss": 0.1229,
+      "step": 9104
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8684086629001883e-05,
+      "loss": 0.1052,
+      "step": 9105
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8681732580037667e-05,
+      "loss": 0.2183,
+      "step": 9106
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8679378531073447e-05,
+      "loss": 0.0389,
+      "step": 9107
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8677024482109228e-05,
+      "loss": 0.1077,
+      "step": 9108
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8674670433145012e-05,
+      "loss": 0.1042,
+      "step": 9109
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8672316384180792e-05,
+      "loss": 0.1705,
+      "step": 9110
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8669962335216573e-05,
+      "loss": 0.0282,
+      "step": 9111
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8667608286252357e-05,
+      "loss": 0.0363,
+      "step": 9112
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8665254237288137e-05,
+      "loss": 0.0625,
+      "step": 9113
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.866290018832392e-05,
+      "loss": 0.1336,
+      "step": 9114
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8660546139359702e-05,
+      "loss": 0.1714,
+      "step": 9115
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.865819209039548e-05,
+      "loss": 0.0844,
+      "step": 9116
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8655838041431266e-05,
+      "loss": 0.0773,
+      "step": 9117
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8653483992467044e-05,
+      "loss": 0.0342,
+      "step": 9118
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8651129943502824e-05,
+      "loss": 0.0526,
+      "step": 9119
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.864877589453861e-05,
+      "loss": 0.0625,
+      "step": 9120
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.864642184557439e-05,
+      "loss": 0.0452,
+      "step": 9121
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.864406779661017e-05,
+      "loss": 0.1164,
+      "step": 9122
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8641713747645953e-05,
+      "loss": 0.1728,
+      "step": 9123
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8639359698681734e-05,
+      "loss": 0.1186,
+      "step": 9124
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8637005649717514e-05,
+      "loss": 0.1212,
+      "step": 9125
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8634651600753298e-05,
+      "loss": 0.0705,
+      "step": 9126
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.863229755178908e-05,
+      "loss": 0.099,
+      "step": 9127
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.862994350282486e-05,
+      "loss": 0.0771,
+      "step": 9128
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8627589453860643e-05,
+      "loss": 0.1232,
+      "step": 9129
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8625235404896424e-05,
+      "loss": 0.1471,
+      "step": 9130
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8622881355932207e-05,
+      "loss": 0.1054,
+      "step": 9131
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8620527306967988e-05,
+      "loss": 0.1467,
+      "step": 9132
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8618173258003765e-05,
+      "loss": 0.0497,
+      "step": 9133
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8615819209039552e-05,
+      "loss": 0.0658,
+      "step": 9134
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.861346516007533e-05,
+      "loss": 0.0374,
+      "step": 9135
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.861111111111111e-05,
+      "loss": 0.1287,
+      "step": 9136
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8608757062146897e-05,
+      "loss": 0.0425,
+      "step": 9137
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8606403013182675e-05,
+      "loss": 0.0281,
+      "step": 9138
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8604048964218455e-05,
+      "loss": 0.1912,
+      "step": 9139
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.860169491525424e-05,
+      "loss": 0.0669,
+      "step": 9140
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.859934086629002e-05,
+      "loss": 0.0748,
+      "step": 9141
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.85969868173258e-05,
+      "loss": 0.1056,
+      "step": 9142
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8594632768361584e-05,
+      "loss": 0.0459,
+      "step": 9143
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8592278719397365e-05,
+      "loss": 0.0333,
+      "step": 9144
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.858992467043315e-05,
+      "loss": 0.0758,
+      "step": 9145
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.858757062146893e-05,
+      "loss": 0.0901,
+      "step": 9146
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.858521657250471e-05,
+      "loss": 0.0912,
+      "step": 9147
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8582862523540493e-05,
+      "loss": 0.1864,
+      "step": 9148
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8580508474576274e-05,
+      "loss": 0.1667,
+      "step": 9149
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.857815442561205e-05,
+      "loss": 0.0707,
+      "step": 9150
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.857580037664784e-05,
+      "loss": 0.0194,
+      "step": 9151
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8573446327683616e-05,
+      "loss": 0.1916,
+      "step": 9152
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8571092278719396e-05,
+      "loss": 0.1002,
+      "step": 9153
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8568738229755183e-05,
+      "loss": 0.0374,
+      "step": 9154
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.856638418079096e-05,
+      "loss": 0.0769,
+      "step": 9155
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.856403013182674e-05,
+      "loss": 0.113,
+      "step": 9156
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8561676082862525e-05,
+      "loss": 0.0968,
+      "step": 9157
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8559322033898306e-05,
+      "loss": 0.1212,
+      "step": 9158
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8556967984934086e-05,
+      "loss": 0.089,
+      "step": 9159
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.855461393596987e-05,
+      "loss": 0.0511,
+      "step": 9160
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.855225988700565e-05,
+      "loss": 0.1707,
+      "step": 9161
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8549905838041435e-05,
+      "loss": 0.0788,
+      "step": 9162
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8547551789077215e-05,
+      "loss": 0.0656,
+      "step": 9163
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8545197740112996e-05,
+      "loss": 0.1018,
+      "step": 9164
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.854284369114878e-05,
+      "loss": 0.2561,
+      "step": 9165
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.854048964218456e-05,
+      "loss": 0.1139,
+      "step": 9166
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8538135593220337e-05,
+      "loss": 0.1324,
+      "step": 9167
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8535781544256125e-05,
+      "loss": 0.1256,
+      "step": 9168
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8533427495291905e-05,
+      "loss": 0.0524,
+      "step": 9169
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8531073446327682e-05,
+      "loss": 0.1447,
+      "step": 9170
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.852871939736347e-05,
+      "loss": 0.1064,
+      "step": 9171
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8526365348399247e-05,
+      "loss": 0.1144,
+      "step": 9172
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8524011299435027e-05,
+      "loss": 0.1078,
+      "step": 9173
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.852165725047081e-05,
+      "loss": 0.1759,
+      "step": 9174
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.851930320150659e-05,
+      "loss": 0.1588,
+      "step": 9175
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8516949152542372e-05,
+      "loss": 0.1178,
+      "step": 9176
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8514595103578156e-05,
+      "loss": 0.0621,
+      "step": 9177
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8512241054613937e-05,
+      "loss": 0.1484,
+      "step": 9178
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.850988700564972e-05,
+      "loss": 0.1154,
+      "step": 9179
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.85075329566855e-05,
+      "loss": 0.0395,
+      "step": 9180
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.850517890772128e-05,
+      "loss": 0.1376,
+      "step": 9181
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8502824858757066e-05,
+      "loss": 0.0452,
+      "step": 9182
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8500470809792846e-05,
+      "loss": 0.1125,
+      "step": 9183
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8498116760828623e-05,
+      "loss": 0.1837,
+      "step": 9184
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.849576271186441e-05,
+      "loss": 0.0249,
+      "step": 9185
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.849340866290019e-05,
+      "loss": 0.0436,
+      "step": 9186
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8491054613935968e-05,
+      "loss": 0.0306,
+      "step": 9187
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8488700564971756e-05,
+      "loss": 0.0747,
+      "step": 9188
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8486346516007533e-05,
+      "loss": 0.048,
+      "step": 9189
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8483992467043313e-05,
+      "loss": 0.1472,
+      "step": 9190
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8481638418079097e-05,
+      "loss": 0.097,
+      "step": 9191
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8479284369114878e-05,
+      "loss": 0.0163,
+      "step": 9192
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.847693032015066e-05,
+      "loss": 0.0326,
+      "step": 9193
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8474576271186442e-05,
+      "loss": 0.0898,
+      "step": 9194
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8472222222222223e-05,
+      "loss": 0.1224,
+      "step": 9195
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8469868173258007e-05,
+      "loss": 0.1319,
+      "step": 9196
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8467514124293787e-05,
+      "loss": 0.0847,
+      "step": 9197
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8465160075329568e-05,
+      "loss": 0.0532,
+      "step": 9198
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.846280602636535e-05,
+      "loss": 0.0953,
+      "step": 9199
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8460451977401132e-05,
+      "loss": 0.1134,
+      "step": 9200
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.845809792843691e-05,
+      "loss": 0.1383,
+      "step": 9201
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8455743879472697e-05,
+      "loss": 0.0838,
+      "step": 9202
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8453389830508477e-05,
+      "loss": 0.052,
+      "step": 9203
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8451035781544254e-05,
+      "loss": 0.0847,
+      "step": 9204
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.844868173258004e-05,
+      "loss": 0.0543,
+      "step": 9205
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.844632768361582e-05,
+      "loss": 0.0585,
+      "step": 9206
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.84439736346516e-05,
+      "loss": 0.2064,
+      "step": 9207
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8441619585687383e-05,
+      "loss": 0.0798,
+      "step": 9208
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8439265536723164e-05,
+      "loss": 0.0889,
+      "step": 9209
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8436911487758948e-05,
+      "loss": 0.1241,
+      "step": 9210
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8434557438794728e-05,
+      "loss": 0.1122,
+      "step": 9211
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.843220338983051e-05,
+      "loss": 0.0195,
+      "step": 9212
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8429849340866293e-05,
+      "loss": 0.2092,
+      "step": 9213
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8427495291902073e-05,
+      "loss": 0.138,
+      "step": 9214
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8425141242937854e-05,
+      "loss": 0.1254,
+      "step": 9215
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8422787193973638e-05,
+      "loss": 0.1167,
+      "step": 9216
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8420433145009418e-05,
+      "loss": 0.1062,
+      "step": 9217
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8418079096045195e-05,
+      "loss": 0.2817,
+      "step": 9218
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8415725047080983e-05,
+      "loss": 0.104,
+      "step": 9219
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8413370998116763e-05,
+      "loss": 0.0635,
+      "step": 9220
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.841101694915254e-05,
+      "loss": 0.0465,
+      "step": 9221
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8408662900188328e-05,
+      "loss": 0.1744,
+      "step": 9222
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8406308851224105e-05,
+      "loss": 0.1713,
+      "step": 9223
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8403954802259892e-05,
+      "loss": 0.1169,
+      "step": 9224
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.840160075329567e-05,
+      "loss": 0.0754,
+      "step": 9225
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.839924670433145e-05,
+      "loss": 0.1157,
+      "step": 9226
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8396892655367234e-05,
+      "loss": 0.1077,
+      "step": 9227
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8394538606403014e-05,
+      "loss": 0.0276,
+      "step": 9228
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8392184557438795e-05,
+      "loss": 0.2404,
+      "step": 9229
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.838983050847458e-05,
+      "loss": 0.1517,
+      "step": 9230
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.838747645951036e-05,
+      "loss": 0.1552,
+      "step": 9231
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.838512241054614e-05,
+      "loss": 0.0733,
+      "step": 9232
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8382768361581924e-05,
+      "loss": 0.1011,
+      "step": 9233
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8380414312617704e-05,
+      "loss": 0.1082,
+      "step": 9234
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.837806026365348e-05,
+      "loss": 0.2769,
+      "step": 9235
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.837570621468927e-05,
+      "loss": 0.1222,
+      "step": 9236
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.837335216572505e-05,
+      "loss": 0.0824,
+      "step": 9237
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8370998116760826e-05,
+      "loss": 0.0662,
+      "step": 9238
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8368644067796614e-05,
+      "loss": 0.0865,
+      "step": 9239
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.836629001883239e-05,
+      "loss": 0.0794,
+      "step": 9240
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8363935969868178e-05,
+      "loss": 0.0305,
+      "step": 9241
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8361581920903955e-05,
+      "loss": 0.1162,
+      "step": 9242
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8359227871939736e-05,
+      "loss": 0.0609,
+      "step": 9243
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.835687382297552e-05,
+      "loss": 0.0185,
+      "step": 9244
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.83545197740113e-05,
+      "loss": 0.069,
+      "step": 9245
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.835216572504708e-05,
+      "loss": 0.1964,
+      "step": 9246
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8349811676082865e-05,
+      "loss": 0.0727,
+      "step": 9247
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8347457627118645e-05,
+      "loss": 0.0258,
+      "step": 9248
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8345103578154426e-05,
+      "loss": 0.1261,
+      "step": 9249
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.834274952919021e-05,
+      "loss": 0.0448,
+      "step": 9250
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.834039548022599e-05,
+      "loss": 0.1134,
+      "step": 9251
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.833804143126177e-05,
+      "loss": 0.1141,
+      "step": 9252
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8335687382297555e-05,
+      "loss": 0.0722,
+      "step": 9253
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8333333333333335e-05,
+      "loss": 0.1226,
+      "step": 9254
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8330979284369113e-05,
+      "loss": 0.1243,
+      "step": 9255
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.83286252354049e-05,
+      "loss": 0.0163,
+      "step": 9256
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8326271186440677e-05,
+      "loss": 0.082,
+      "step": 9257
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8323917137476464e-05,
+      "loss": 0.1985,
+      "step": 9258
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.832156308851224e-05,
+      "loss": 0.1573,
+      "step": 9259
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8319209039548022e-05,
+      "loss": 0.164,
+      "step": 9260
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.831685499058381e-05,
+      "loss": 0.0749,
+      "step": 9261
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8314500941619586e-05,
+      "loss": 0.0968,
+      "step": 9262
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8312146892655367e-05,
+      "loss": 0.2215,
+      "step": 9263
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.830979284369115e-05,
+      "loss": 0.1752,
+      "step": 9264
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.830743879472693e-05,
+      "loss": 0.175,
+      "step": 9265
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8305084745762712e-05,
+      "loss": 0.1309,
+      "step": 9266
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8302730696798496e-05,
+      "loss": 0.0684,
+      "step": 9267
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8300376647834276e-05,
+      "loss": 0.0751,
+      "step": 9268
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8298022598870057e-05,
+      "loss": 0.1296,
+      "step": 9269
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.829566854990584e-05,
+      "loss": 0.0991,
+      "step": 9270
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.829331450094162e-05,
+      "loss": 0.0658,
+      "step": 9271
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8290960451977405e-05,
+      "loss": 0.0614,
+      "step": 9272
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8288606403013186e-05,
+      "loss": 0.1267,
+      "step": 9273
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8286252354048963e-05,
+      "loss": 0.116,
+      "step": 9274
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.828389830508475e-05,
+      "loss": 0.1105,
+      "step": 9275
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8281544256120527e-05,
+      "loss": 0.1146,
+      "step": 9276
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8279190207156308e-05,
+      "loss": 0.066,
+      "step": 9277
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8276836158192095e-05,
+      "loss": 0.1072,
+      "step": 9278
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8274482109227872e-05,
+      "loss": 0.1542,
+      "step": 9279
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8272128060263653e-05,
+      "loss": 0.0789,
+      "step": 9280
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8269774011299437e-05,
+      "loss": 0.0668,
+      "step": 9281
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8267419962335217e-05,
+      "loss": 0.1377,
+      "step": 9282
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8265065913370998e-05,
+      "loss": 0.0503,
+      "step": 9283
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8262711864406782e-05,
+      "loss": 0.1545,
+      "step": 9284
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8260357815442562e-05,
+      "loss": 0.067,
+      "step": 9285
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8258003766478343e-05,
+      "loss": 0.0847,
+      "step": 9286
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8255649717514127e-05,
+      "loss": 0.09,
+      "step": 9287
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8253295668549907e-05,
+      "loss": 0.0525,
+      "step": 9288
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.825094161958569e-05,
+      "loss": 0.1076,
+      "step": 9289
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8248587570621472e-05,
+      "loss": 0.1218,
+      "step": 9290
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.824623352165725e-05,
+      "loss": 0.1013,
+      "step": 9291
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8243879472693036e-05,
+      "loss": 0.0794,
+      "step": 9292
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8241525423728814e-05,
+      "loss": 0.1424,
+      "step": 9293
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8239171374764594e-05,
+      "loss": 0.0399,
+      "step": 9294
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.823681732580038e-05,
+      "loss": 0.019,
+      "step": 9295
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.823446327683616e-05,
+      "loss": 0.0588,
+      "step": 9296
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.823210922787194e-05,
+      "loss": 0.0619,
+      "step": 9297
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8229755178907723e-05,
+      "loss": 0.1189,
+      "step": 9298
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8227401129943504e-05,
+      "loss": 0.061,
+      "step": 9299
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8225047080979284e-05,
+      "loss": 0.0697,
+      "step": 9300
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8222693032015068e-05,
+      "loss": 0.1237,
+      "step": 9301
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.822033898305085e-05,
+      "loss": 0.0873,
+      "step": 9302
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8217984934086632e-05,
+      "loss": 0.1059,
+      "step": 9303
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8215630885122413e-05,
+      "loss": 0.1403,
+      "step": 9304
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8213276836158194e-05,
+      "loss": 0.0721,
+      "step": 9305
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8210922787193977e-05,
+      "loss": 0.1194,
+      "step": 9306
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8208568738229758e-05,
+      "loss": 0.0928,
+      "step": 9307
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8206214689265535e-05,
+      "loss": 0.0682,
+      "step": 9308
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8203860640301322e-05,
+      "loss": 0.1597,
+      "step": 9309
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.82015065913371e-05,
+      "loss": 0.1534,
+      "step": 9310
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.819915254237288e-05,
+      "loss": 0.0394,
+      "step": 9311
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8196798493408667e-05,
+      "loss": 0.0192,
+      "step": 9312
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8194444444444445e-05,
+      "loss": 0.0422,
+      "step": 9313
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8192090395480225e-05,
+      "loss": 0.121,
+      "step": 9314
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.818973634651601e-05,
+      "loss": 0.0944,
+      "step": 9315
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.818738229755179e-05,
+      "loss": 0.0518,
+      "step": 9316
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.818502824858757e-05,
+      "loss": 0.0749,
+      "step": 9317
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8182674199623354e-05,
+      "loss": 0.0866,
+      "step": 9318
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8180320150659135e-05,
+      "loss": 0.026,
+      "step": 9319
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.817796610169492e-05,
+      "loss": 0.1485,
+      "step": 9320
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.81756120527307e-05,
+      "loss": 0.0109,
+      "step": 9321
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.817325800376648e-05,
+      "loss": 0.0704,
+      "step": 9322
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8170903954802264e-05,
+      "loss": 0.1759,
+      "step": 9323
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8168549905838044e-05,
+      "loss": 0.0293,
+      "step": 9324
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.816619585687382e-05,
+      "loss": 0.1569,
+      "step": 9325
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.816384180790961e-05,
+      "loss": 0.1013,
+      "step": 9326
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.816148775894539e-05,
+      "loss": 0.1057,
+      "step": 9327
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8159133709981166e-05,
+      "loss": 0.0924,
+      "step": 9328
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8156779661016954e-05,
+      "loss": 0.0598,
+      "step": 9329
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.815442561205273e-05,
+      "loss": 0.1238,
+      "step": 9330
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.815207156308851e-05,
+      "loss": 0.2567,
+      "step": 9331
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8149717514124295e-05,
+      "loss": 0.0312,
+      "step": 9332
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8147363465160076e-05,
+      "loss": 0.1186,
+      "step": 9333
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8145009416195856e-05,
+      "loss": 0.0877,
+      "step": 9334
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.814265536723164e-05,
+      "loss": 0.086,
+      "step": 9335
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.814030131826742e-05,
+      "loss": 0.1547,
+      "step": 9336
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8137947269303205e-05,
+      "loss": 0.0389,
+      "step": 9337
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8135593220338985e-05,
+      "loss": 0.1478,
+      "step": 9338
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8133239171374766e-05,
+      "loss": 0.0436,
+      "step": 9339
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.813088512241055e-05,
+      "loss": 0.1147,
+      "step": 9340
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.812853107344633e-05,
+      "loss": 0.1049,
+      "step": 9341
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8126177024482107e-05,
+      "loss": 0.1024,
+      "step": 9342
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8123822975517895e-05,
+      "loss": 0.0858,
+      "step": 9343
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8121468926553675e-05,
+      "loss": 0.2249,
+      "step": 9344
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8119114877589452e-05,
+      "loss": 0.0199,
+      "step": 9345
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.811676082862524e-05,
+      "loss": 0.1636,
+      "step": 9346
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8114406779661017e-05,
+      "loss": 0.0642,
+      "step": 9347
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8112052730696797e-05,
+      "loss": 0.0618,
+      "step": 9348
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.810969868173258e-05,
+      "loss": 0.0904,
+      "step": 9349
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8107344632768362e-05,
+      "loss": 0.2192,
+      "step": 9350
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8104990583804146e-05,
+      "loss": 0.2432,
+      "step": 9351
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8102636534839926e-05,
+      "loss": 0.0701,
+      "step": 9352
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8100282485875707e-05,
+      "loss": 0.1815,
+      "step": 9353
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.809792843691149e-05,
+      "loss": 0.1099,
+      "step": 9354
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.809557438794727e-05,
+      "loss": 0.0822,
+      "step": 9355
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8093220338983052e-05,
+      "loss": 0.142,
+      "step": 9356
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8090866290018836e-05,
+      "loss": 0.1083,
+      "step": 9357
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8088512241054616e-05,
+      "loss": 0.0543,
+      "step": 9358
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8086158192090393e-05,
+      "loss": 0.1064,
+      "step": 9359
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.808380414312618e-05,
+      "loss": 0.2286,
+      "step": 9360
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.808145009416196e-05,
+      "loss": 0.1181,
+      "step": 9361
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.807909604519774e-05,
+      "loss": 0.0301,
+      "step": 9362
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8076741996233526e-05,
+      "loss": 0.1446,
+      "step": 9363
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8074387947269303e-05,
+      "loss": 0.1906,
+      "step": 9364
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8072033898305083e-05,
+      "loss": 0.0224,
+      "step": 9365
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8069679849340867e-05,
+      "loss": 0.1176,
+      "step": 9366
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8067325800376648e-05,
+      "loss": 0.1096,
+      "step": 9367
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8064971751412432e-05,
+      "loss": 0.033,
+      "step": 9368
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8062617702448212e-05,
+      "loss": 0.097,
+      "step": 9369
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8060263653483993e-05,
+      "loss": 0.1563,
+      "step": 9370
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8057909604519777e-05,
+      "loss": 0.0968,
+      "step": 9371
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8055555555555557e-05,
+      "loss": 0.0958,
+      "step": 9372
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8053201506591338e-05,
+      "loss": 0.0525,
+      "step": 9373
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8050847457627122e-05,
+      "loss": 0.0667,
+      "step": 9374
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8048493408662902e-05,
+      "loss": 0.0571,
+      "step": 9375
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.804613935969868e-05,
+      "loss": 0.1429,
+      "step": 9376
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8043785310734467e-05,
+      "loss": 0.0596,
+      "step": 9377
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8041431261770247e-05,
+      "loss": 0.1572,
+      "step": 9378
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.8039077212806024e-05,
+      "loss": 0.076,
+      "step": 9379
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.803672316384181e-05,
+      "loss": 0.0822,
+      "step": 9380
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.803436911487759e-05,
+      "loss": 0.0721,
+      "step": 9381
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.803201506591337e-05,
+      "loss": 0.0249,
+      "step": 9382
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.8029661016949153e-05,
+      "loss": 0.1062,
+      "step": 9383
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.8027306967984934e-05,
+      "loss": 0.0667,
+      "step": 9384
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.8024952919020718e-05,
+      "loss": 0.1577,
+      "step": 9385
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.8022598870056498e-05,
+      "loss": 0.0751,
+      "step": 9386
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.802024482109228e-05,
+      "loss": 0.1817,
+      "step": 9387
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.8017890772128063e-05,
+      "loss": 0.1329,
+      "step": 9388
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.8015536723163843e-05,
+      "loss": 0.088,
+      "step": 9389
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.8013182674199624e-05,
+      "loss": 0.204,
+      "step": 9390
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.8010828625235408e-05,
+      "loss": 0.0861,
+      "step": 9391
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.8008474576271188e-05,
+      "loss": 0.2703,
+      "step": 9392
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.8006120527306965e-05,
+      "loss": 0.051,
+      "step": 9393
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.8003766478342753e-05,
+      "loss": 0.1408,
+      "step": 9394
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.8001412429378533e-05,
+      "loss": 0.0645,
+      "step": 9395
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.799905838041431e-05,
+      "loss": 0.0557,
+      "step": 9396
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7996704331450098e-05,
+      "loss": 0.0185,
+      "step": 9397
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7994350282485875e-05,
+      "loss": 0.0523,
+      "step": 9398
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7991996233521662e-05,
+      "loss": 0.0712,
+      "step": 9399
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.798964218455744e-05,
+      "loss": 0.1124,
+      "step": 9400
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.798728813559322e-05,
+      "loss": 0.1113,
+      "step": 9401
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7984934086629004e-05,
+      "loss": 0.14,
+      "step": 9402
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7982580037664784e-05,
+      "loss": 0.1093,
+      "step": 9403
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7980225988700565e-05,
+      "loss": 0.0765,
+      "step": 9404
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.797787193973635e-05,
+      "loss": 0.0352,
+      "step": 9405
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.797551789077213e-05,
+      "loss": 0.1423,
+      "step": 9406
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.797316384180791e-05,
+      "loss": 0.1285,
+      "step": 9407
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7970809792843694e-05,
+      "loss": 0.0966,
+      "step": 9408
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7968455743879474e-05,
+      "loss": 0.1754,
+      "step": 9409
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7966101694915255e-05,
+      "loss": 0.0958,
+      "step": 9410
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.796374764595104e-05,
+      "loss": 0.1064,
+      "step": 9411
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.796139359698682e-05,
+      "loss": 0.0956,
+      "step": 9412
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7959039548022596e-05,
+      "loss": 0.0432,
+      "step": 9413
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7956685499058384e-05,
+      "loss": 0.0532,
+      "step": 9414
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.795433145009416e-05,
+      "loss": 0.0499,
+      "step": 9415
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7951977401129948e-05,
+      "loss": 0.1083,
+      "step": 9416
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7949623352165725e-05,
+      "loss": 0.0949,
+      "step": 9417
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7947269303201506e-05,
+      "loss": 0.0292,
+      "step": 9418
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7944915254237293e-05,
+      "loss": 0.0326,
+      "step": 9419
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.794256120527307e-05,
+      "loss": 0.0296,
+      "step": 9420
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.794020715630885e-05,
+      "loss": 0.0854,
+      "step": 9421
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7937853107344635e-05,
+      "loss": 0.0979,
+      "step": 9422
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7935499058380415e-05,
+      "loss": 0.2014,
+      "step": 9423
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7933145009416196e-05,
+      "loss": 0.1032,
+      "step": 9424
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.793079096045198e-05,
+      "loss": 0.0845,
+      "step": 9425
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.792843691148776e-05,
+      "loss": 0.1377,
+      "step": 9426
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.792608286252354e-05,
+      "loss": 0.104,
+      "step": 9427
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7923728813559325e-05,
+      "loss": 0.0312,
+      "step": 9428
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7921374764595105e-05,
+      "loss": 0.1148,
+      "step": 9429
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.791902071563089e-05,
+      "loss": 0.0315,
+      "step": 9430
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.791666666666667e-05,
+      "loss": 0.2161,
+      "step": 9431
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7914312617702447e-05,
+      "loss": 0.0389,
+      "step": 9432
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7911958568738234e-05,
+      "loss": 0.0474,
+      "step": 9433
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.790960451977401e-05,
+      "loss": 0.0805,
+      "step": 9434
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7907250470809792e-05,
+      "loss": 0.1796,
+      "step": 9435
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.790489642184558e-05,
+      "loss": 0.1623,
+      "step": 9436
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7902542372881356e-05,
+      "loss": 0.12,
+      "step": 9437
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7900188323917137e-05,
+      "loss": 0.0936,
+      "step": 9438
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.789783427495292e-05,
+      "loss": 0.1139,
+      "step": 9439
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.78954802259887e-05,
+      "loss": 0.16,
+      "step": 9440
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7893126177024482e-05,
+      "loss": 0.0776,
+      "step": 9441
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7890772128060266e-05,
+      "loss": 0.074,
+      "step": 9442
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7888418079096046e-05,
+      "loss": 0.0982,
+      "step": 9443
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7886064030131827e-05,
+      "loss": 0.1617,
+      "step": 9444
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.788370998116761e-05,
+      "loss": 0.1873,
+      "step": 9445
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.788135593220339e-05,
+      "loss": 0.0536,
+      "step": 9446
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7879001883239175e-05,
+      "loss": 0.0631,
+      "step": 9447
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7876647834274956e-05,
+      "loss": 0.0634,
+      "step": 9448
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7874293785310733e-05,
+      "loss": 0.0767,
+      "step": 9449
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.787193973634652e-05,
+      "loss": 0.0821,
+      "step": 9450
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7869585687382298e-05,
+      "loss": 0.0348,
+      "step": 9451
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7867231638418078e-05,
+      "loss": 0.0686,
+      "step": 9452
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7864877589453865e-05,
+      "loss": 0.1261,
+      "step": 9453
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7862523540489643e-05,
+      "loss": 0.0313,
+      "step": 9454
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7860169491525423e-05,
+      "loss": 0.316,
+      "step": 9455
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7857815442561207e-05,
+      "loss": 0.186,
+      "step": 9456
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7855461393596988e-05,
+      "loss": 0.2483,
+      "step": 9457
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7853107344632768e-05,
+      "loss": 0.0593,
+      "step": 9458
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7850753295668552e-05,
+      "loss": 0.0762,
+      "step": 9459
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7848399246704333e-05,
+      "loss": 0.0612,
+      "step": 9460
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7846045197740113e-05,
+      "loss": 0.1032,
+      "step": 9461
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7843691148775897e-05,
+      "loss": 0.0633,
+      "step": 9462
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7841337099811678e-05,
+      "loss": 0.1885,
+      "step": 9463
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.783898305084746e-05,
+      "loss": 0.0521,
+      "step": 9464
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7836629001883242e-05,
+      "loss": 0.058,
+      "step": 9465
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.783427495291902e-05,
+      "loss": 0.095,
+      "step": 9466
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7831920903954806e-05,
+      "loss": 0.1971,
+      "step": 9467
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7829566854990584e-05,
+      "loss": 0.1382,
+      "step": 9468
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7827212806026364e-05,
+      "loss": 0.2564,
+      "step": 9469
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.782485875706215e-05,
+      "loss": 0.1242,
+      "step": 9470
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.782250470809793e-05,
+      "loss": 0.1021,
+      "step": 9471
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.782015065913371e-05,
+      "loss": 0.0848,
+      "step": 9472
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7817796610169493e-05,
+      "loss": 0.0621,
+      "step": 9473
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7815442561205274e-05,
+      "loss": 0.1279,
+      "step": 9474
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7813088512241054e-05,
+      "loss": 0.0572,
+      "step": 9475
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7810734463276838e-05,
+      "loss": 0.1225,
+      "step": 9476
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.780838041431262e-05,
+      "loss": 0.0478,
+      "step": 9477
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7806026365348403e-05,
+      "loss": 0.128,
+      "step": 9478
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7803672316384183e-05,
+      "loss": 0.1135,
+      "step": 9479
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7801318267419964e-05,
+      "loss": 0.0573,
+      "step": 9480
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7798964218455748e-05,
+      "loss": 0.0486,
+      "step": 9481
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7796610169491528e-05,
+      "loss": 0.1115,
+      "step": 9482
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7794256120527305e-05,
+      "loss": 0.0792,
+      "step": 9483
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7791902071563093e-05,
+      "loss": 0.0922,
+      "step": 9484
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7789548022598873e-05,
+      "loss": 0.0223,
+      "step": 9485
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.778719397363465e-05,
+      "loss": 0.07,
+      "step": 9486
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7784839924670437e-05,
+      "loss": 0.0386,
+      "step": 9487
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7782485875706215e-05,
+      "loss": 0.0535,
+      "step": 9488
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7780131826741995e-05,
+      "loss": 0.1974,
+      "step": 9489
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.777777777777778e-05,
+      "loss": 0.0543,
+      "step": 9490
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.777542372881356e-05,
+      "loss": 0.1966,
+      "step": 9491
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.777306967984934e-05,
+      "loss": 0.1848,
+      "step": 9492
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7770715630885124e-05,
+      "loss": 0.0517,
+      "step": 9493
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7768361581920905e-05,
+      "loss": 0.055,
+      "step": 9494
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.776600753295669e-05,
+      "loss": 0.0373,
+      "step": 9495
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.776365348399247e-05,
+      "loss": 0.1469,
+      "step": 9496
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.776129943502825e-05,
+      "loss": 0.1065,
+      "step": 9497
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7758945386064034e-05,
+      "loss": 0.1287,
+      "step": 9498
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7756591337099814e-05,
+      "loss": 0.057,
+      "step": 9499
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.775423728813559e-05,
+      "loss": 0.076,
+      "step": 9500
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.775188323917138e-05,
+      "loss": 0.0777,
+      "step": 9501
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.774952919020716e-05,
+      "loss": 0.1433,
+      "step": 9502
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7747175141242936e-05,
+      "loss": 0.1295,
+      "step": 9503
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7744821092278724e-05,
+      "loss": 0.099,
+      "step": 9504
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.77424670433145e-05,
+      "loss": 0.1388,
+      "step": 9505
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.774011299435028e-05,
+      "loss": 0.0545,
+      "step": 9506
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7737758945386065e-05,
+      "loss": 0.1862,
+      "step": 9507
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7735404896421846e-05,
+      "loss": 0.1455,
+      "step": 9508
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.773305084745763e-05,
+      "loss": 0.0496,
+      "step": 9509
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.773069679849341e-05,
+      "loss": 0.1329,
+      "step": 9510
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.772834274952919e-05,
+      "loss": 0.0475,
+      "step": 9511
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7725988700564975e-05,
+      "loss": 0.0576,
+      "step": 9512
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7723634651600755e-05,
+      "loss": 0.0607,
+      "step": 9513
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7721280602636536e-05,
+      "loss": 0.0349,
+      "step": 9514
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.771892655367232e-05,
+      "loss": 0.1902,
+      "step": 9515
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.77165725047081e-05,
+      "loss": 0.0632,
+      "step": 9516
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7714218455743877e-05,
+      "loss": 0.017,
+      "step": 9517
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7711864406779665e-05,
+      "loss": 0.0877,
+      "step": 9518
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7709510357815445e-05,
+      "loss": 0.0423,
+      "step": 9519
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7707156308851222e-05,
+      "loss": 0.0864,
+      "step": 9520
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.770480225988701e-05,
+      "loss": 0.1179,
+      "step": 9521
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7702448210922787e-05,
+      "loss": 0.0837,
+      "step": 9522
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7700094161958567e-05,
+      "loss": 0.0301,
+      "step": 9523
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.769774011299435e-05,
+      "loss": 0.0714,
+      "step": 9524
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7695386064030132e-05,
+      "loss": 0.1974,
+      "step": 9525
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7693032015065916e-05,
+      "loss": 0.0499,
+      "step": 9526
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7690677966101696e-05,
+      "loss": 0.0773,
+      "step": 9527
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7688323917137477e-05,
+      "loss": 0.1356,
+      "step": 9528
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.768596986817326e-05,
+      "loss": 0.239,
+      "step": 9529
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.768361581920904e-05,
+      "loss": 0.0616,
+      "step": 9530
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7681261770244822e-05,
+      "loss": 0.1534,
+      "step": 9531
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7678907721280606e-05,
+      "loss": 0.0762,
+      "step": 9532
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7676553672316386e-05,
+      "loss": 0.0362,
+      "step": 9533
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7674199623352163e-05,
+      "loss": 0.1582,
+      "step": 9534
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.767184557438795e-05,
+      "loss": 0.0676,
+      "step": 9535
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.766949152542373e-05,
+      "loss": 0.0815,
+      "step": 9536
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.766713747645951e-05,
+      "loss": 0.055,
+      "step": 9537
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7664783427495296e-05,
+      "loss": 0.1993,
+      "step": 9538
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7662429378531073e-05,
+      "loss": 0.0864,
+      "step": 9539
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7660075329566853e-05,
+      "loss": 0.0344,
+      "step": 9540
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7657721280602637e-05,
+      "loss": 0.0884,
+      "step": 9541
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7655367231638418e-05,
+      "loss": 0.1313,
+      "step": 9542
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7653013182674202e-05,
+      "loss": 0.1997,
+      "step": 9543
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7650659133709982e-05,
+      "loss": 0.0935,
+      "step": 9544
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7648305084745763e-05,
+      "loss": 0.0735,
+      "step": 9545
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7645951035781547e-05,
+      "loss": 0.0792,
+      "step": 9546
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7643596986817327e-05,
+      "loss": 0.1218,
+      "step": 9547
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7641242937853108e-05,
+      "loss": 0.0447,
+      "step": 9548
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7638888888888892e-05,
+      "loss": 0.0878,
+      "step": 9549
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7636534839924672e-05,
+      "loss": 0.0123,
+      "step": 9550
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.763418079096045e-05,
+      "loss": 0.0166,
+      "step": 9551
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7631826741996237e-05,
+      "loss": 0.097,
+      "step": 9552
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7629472693032017e-05,
+      "loss": 0.0587,
+      "step": 9553
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7627118644067794e-05,
+      "loss": 0.0822,
+      "step": 9554
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7624764595103582e-05,
+      "loss": 0.0123,
+      "step": 9555
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.762241054613936e-05,
+      "loss": 0.0353,
+      "step": 9556
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7620056497175146e-05,
+      "loss": 0.1361,
+      "step": 9557
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7617702448210923e-05,
+      "loss": 0.0243,
+      "step": 9558
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7615348399246704e-05,
+      "loss": 0.0419,
+      "step": 9559
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7612994350282488e-05,
+      "loss": 0.0843,
+      "step": 9560
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.761064030131827e-05,
+      "loss": 0.0862,
+      "step": 9561
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.760828625235405e-05,
+      "loss": 0.0767,
+      "step": 9562
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7605932203389833e-05,
+      "loss": 0.1955,
+      "step": 9563
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7603578154425613e-05,
+      "loss": 0.1979,
+      "step": 9564
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7601224105461394e-05,
+      "loss": 0.0913,
+      "step": 9565
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7598870056497178e-05,
+      "loss": 0.0855,
+      "step": 9566
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.759651600753296e-05,
+      "loss": 0.1642,
+      "step": 9567
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.759416195856874e-05,
+      "loss": 0.1949,
+      "step": 9568
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7591807909604523e-05,
+      "loss": 0.1175,
+      "step": 9569
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7589453860640303e-05,
+      "loss": 0.0253,
+      "step": 9570
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.758709981167608e-05,
+      "loss": 0.0307,
+      "step": 9571
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7584745762711868e-05,
+      "loss": 0.0236,
+      "step": 9572
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7582391713747645e-05,
+      "loss": 0.1394,
+      "step": 9573
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7580037664783432e-05,
+      "loss": 0.1025,
+      "step": 9574
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.757768361581921e-05,
+      "loss": 0.0697,
+      "step": 9575
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.757532956685499e-05,
+      "loss": 0.1433,
+      "step": 9576
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7572975517890777e-05,
+      "loss": 0.1519,
+      "step": 9577
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7570621468926554e-05,
+      "loss": 0.0942,
+      "step": 9578
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7568267419962335e-05,
+      "loss": 0.1619,
+      "step": 9579
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.756591337099812e-05,
+      "loss": 0.1214,
+      "step": 9580
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.75635593220339e-05,
+      "loss": 0.1052,
+      "step": 9581
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.756120527306968e-05,
+      "loss": 0.0291,
+      "step": 9582
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7558851224105464e-05,
+      "loss": 0.1133,
+      "step": 9583
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7556497175141244e-05,
+      "loss": 0.1168,
+      "step": 9584
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7554143126177025e-05,
+      "loss": 0.0907,
+      "step": 9585
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.755178907721281e-05,
+      "loss": 0.1361,
+      "step": 9586
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.754943502824859e-05,
+      "loss": 0.2137,
+      "step": 9587
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7547080979284373e-05,
+      "loss": 0.1409,
+      "step": 9588
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7544726930320154e-05,
+      "loss": 0.1006,
+      "step": 9589
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.754237288135593e-05,
+      "loss": 0.0429,
+      "step": 9590
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.754001883239172e-05,
+      "loss": 0.065,
+      "step": 9591
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7537664783427495e-05,
+      "loss": 0.0648,
+      "step": 9592
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7535310734463276e-05,
+      "loss": 0.077,
+      "step": 9593
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7532956685499063e-05,
+      "loss": 0.0688,
+      "step": 9594
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.753060263653484e-05,
+      "loss": 0.1493,
+      "step": 9595
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.752824858757062e-05,
+      "loss": 0.1265,
+      "step": 9596
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7525894538606405e-05,
+      "loss": 0.1256,
+      "step": 9597
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7523540489642185e-05,
+      "loss": 0.1982,
+      "step": 9598
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7521186440677966e-05,
+      "loss": 0.0312,
+      "step": 9599
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.751883239171375e-05,
+      "loss": 0.1204,
+      "step": 9600
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.751647834274953e-05,
+      "loss": 0.1009,
+      "step": 9601
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.751412429378531e-05,
+      "loss": 0.0871,
+      "step": 9602
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7511770244821095e-05,
+      "loss": 0.1961,
+      "step": 9603
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7509416195856875e-05,
+      "loss": 0.2184,
+      "step": 9604
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.750706214689266e-05,
+      "loss": 0.138,
+      "step": 9605
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.750470809792844e-05,
+      "loss": 0.1089,
+      "step": 9606
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7502354048964217e-05,
+      "loss": 0.0425,
+      "step": 9607
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7500000000000004e-05,
+      "loss": 0.0997,
+      "step": 9608
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.749764595103578e-05,
+      "loss": 0.0835,
+      "step": 9609
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7495291902071562e-05,
+      "loss": 0.055,
+      "step": 9610
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.749293785310735e-05,
+      "loss": 0.1496,
+      "step": 9611
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7490583804143127e-05,
+      "loss": 0.0883,
+      "step": 9612
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7488229755178907e-05,
+      "loss": 0.0417,
+      "step": 9613
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.748587570621469e-05,
+      "loss": 0.1027,
+      "step": 9614
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.748352165725047e-05,
+      "loss": 0.1153,
+      "step": 9615
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7481167608286252e-05,
+      "loss": 0.1332,
+      "step": 9616
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7478813559322036e-05,
+      "loss": 0.1246,
+      "step": 9617
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7476459510357817e-05,
+      "loss": 0.1258,
+      "step": 9618
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7474105461393597e-05,
+      "loss": 0.0283,
+      "step": 9619
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.747175141242938e-05,
+      "loss": 0.1347,
+      "step": 9620
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.746939736346516e-05,
+      "loss": 0.0657,
+      "step": 9621
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7467043314500945e-05,
+      "loss": 0.0224,
+      "step": 9622
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7464689265536726e-05,
+      "loss": 0.1005,
+      "step": 9623
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7462335216572503e-05,
+      "loss": 0.1155,
+      "step": 9624
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.745998116760829e-05,
+      "loss": 0.0588,
+      "step": 9625
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7457627118644068e-05,
+      "loss": 0.0486,
+      "step": 9626
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7455273069679848e-05,
+      "loss": 0.2929,
+      "step": 9627
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7452919020715635e-05,
+      "loss": 0.0195,
+      "step": 9628
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7450564971751413e-05,
+      "loss": 0.0184,
+      "step": 9629
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7448210922787193e-05,
+      "loss": 0.0951,
+      "step": 9630
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7445856873822977e-05,
+      "loss": 0.0744,
+      "step": 9631
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7443502824858758e-05,
+      "loss": 0.041,
+      "step": 9632
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7441148775894538e-05,
+      "loss": 0.056,
+      "step": 9633
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7438794726930322e-05,
+      "loss": 0.0843,
+      "step": 9634
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7436440677966103e-05,
+      "loss": 0.1362,
+      "step": 9635
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7434086629001886e-05,
+      "loss": 0.1008,
+      "step": 9636
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7431732580037667e-05,
+      "loss": 0.15,
+      "step": 9637
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7429378531073448e-05,
+      "loss": 0.0541,
+      "step": 9638
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.742702448210923e-05,
+      "loss": 0.0115,
+      "step": 9639
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7424670433145012e-05,
+      "loss": 0.0983,
+      "step": 9640
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.742231638418079e-05,
+      "loss": 0.1325,
+      "step": 9641
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7419962335216576e-05,
+      "loss": 0.0862,
+      "step": 9642
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7417608286252357e-05,
+      "loss": 0.1954,
+      "step": 9643
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7415254237288134e-05,
+      "loss": 0.0709,
+      "step": 9644
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.741290018832392e-05,
+      "loss": 0.1444,
+      "step": 9645
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.74105461393597e-05,
+      "loss": 0.0621,
+      "step": 9646
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.740819209039548e-05,
+      "loss": 0.0933,
+      "step": 9647
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7405838041431263e-05,
+      "loss": 0.1785,
+      "step": 9648
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7403483992467044e-05,
+      "loss": 0.0954,
+      "step": 9649
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7401129943502824e-05,
+      "loss": 0.0163,
+      "step": 9650
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7398775894538608e-05,
+      "loss": 0.119,
+      "step": 9651
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.739642184557439e-05,
+      "loss": 0.0822,
+      "step": 9652
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7394067796610173e-05,
+      "loss": 0.1189,
+      "step": 9653
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7391713747645953e-05,
+      "loss": 0.0898,
+      "step": 9654
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7389359698681734e-05,
+      "loss": 0.134,
+      "step": 9655
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7387005649717518e-05,
+      "loss": 0.061,
+      "step": 9656
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7384651600753298e-05,
+      "loss": 0.0737,
+      "step": 9657
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7382297551789075e-05,
+      "loss": 0.0165,
+      "step": 9658
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7379943502824863e-05,
+      "loss": 0.0961,
+      "step": 9659
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7377589453860643e-05,
+      "loss": 0.1161,
+      "step": 9660
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.737523540489642e-05,
+      "loss": 0.0915,
+      "step": 9661
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7372881355932208e-05,
+      "loss": 0.1063,
+      "step": 9662
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7370527306967985e-05,
+      "loss": 0.1315,
+      "step": 9663
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7368173258003765e-05,
+      "loss": 0.1391,
+      "step": 9664
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.736581920903955e-05,
+      "loss": 0.1563,
+      "step": 9665
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.736346516007533e-05,
+      "loss": 0.0559,
+      "step": 9666
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7361111111111114e-05,
+      "loss": 0.0825,
+      "step": 9667
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7358757062146894e-05,
+      "loss": 0.2368,
+      "step": 9668
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7356403013182675e-05,
+      "loss": 0.2006,
+      "step": 9669
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.735404896421846e-05,
+      "loss": 0.2088,
+      "step": 9670
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.735169491525424e-05,
+      "loss": 0.0189,
+      "step": 9671
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.734934086629002e-05,
+      "loss": 0.0911,
+      "step": 9672
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7346986817325804e-05,
+      "loss": 0.0818,
+      "step": 9673
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7344632768361584e-05,
+      "loss": 0.1187,
+      "step": 9674
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.734227871939736e-05,
+      "loss": 0.1076,
+      "step": 9675
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.733992467043315e-05,
+      "loss": 0.1751,
+      "step": 9676
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.733757062146893e-05,
+      "loss": 0.0744,
+      "step": 9677
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7335216572504706e-05,
+      "loss": 0.0585,
+      "step": 9678
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7332862523540494e-05,
+      "loss": 0.0486,
+      "step": 9679
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.733050847457627e-05,
+      "loss": 0.0449,
+      "step": 9680
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.732815442561205e-05,
+      "loss": 0.0296,
+      "step": 9681
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7325800376647835e-05,
+      "loss": 0.0865,
+      "step": 9682
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7323446327683616e-05,
+      "loss": 0.0422,
+      "step": 9683
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.73210922787194e-05,
+      "loss": 0.0609,
+      "step": 9684
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.731873822975518e-05,
+      "loss": 0.0429,
+      "step": 9685
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.731638418079096e-05,
+      "loss": 0.148,
+      "step": 9686
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7314030131826745e-05,
+      "loss": 0.0829,
+      "step": 9687
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7311676082862525e-05,
+      "loss": 0.1496,
+      "step": 9688
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7309322033898306e-05,
+      "loss": 0.0242,
+      "step": 9689
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.730696798493409e-05,
+      "loss": 0.1323,
+      "step": 9690
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.730461393596987e-05,
+      "loss": 0.1262,
+      "step": 9691
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7302259887005647e-05,
+      "loss": 0.0517,
+      "step": 9692
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7299905838041435e-05,
+      "loss": 0.0526,
+      "step": 9693
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7297551789077215e-05,
+      "loss": 0.0658,
+      "step": 9694
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7295197740112992e-05,
+      "loss": 0.0457,
+      "step": 9695
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.729284369114878e-05,
+      "loss": 0.2256,
+      "step": 9696
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7290489642184557e-05,
+      "loss": 0.0836,
+      "step": 9697
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7288135593220337e-05,
+      "loss": 0.0333,
+      "step": 9698
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.728578154425612e-05,
+      "loss": 0.0521,
+      "step": 9699
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7283427495291902e-05,
+      "loss": 0.0604,
+      "step": 9700
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7281073446327686e-05,
+      "loss": 0.1025,
+      "step": 9701
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7278719397363466e-05,
+      "loss": 0.054,
+      "step": 9702
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7276365348399247e-05,
+      "loss": 0.0989,
+      "step": 9703
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.727401129943503e-05,
+      "loss": 0.0388,
+      "step": 9704
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.727165725047081e-05,
+      "loss": 0.0122,
+      "step": 9705
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7269303201506592e-05,
+      "loss": 0.0856,
+      "step": 9706
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7266949152542376e-05,
+      "loss": 0.1589,
+      "step": 9707
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7264595103578156e-05,
+      "loss": 0.1098,
+      "step": 9708
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7262241054613933e-05,
+      "loss": 0.1162,
+      "step": 9709
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.725988700564972e-05,
+      "loss": 0.0394,
+      "step": 9710
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.72575329566855e-05,
+      "loss": 0.0722,
+      "step": 9711
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.725517890772128e-05,
+      "loss": 0.1041,
+      "step": 9712
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7252824858757066e-05,
+      "loss": 0.0791,
+      "step": 9713
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7250470809792843e-05,
+      "loss": 0.085,
+      "step": 9714
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.724811676082863e-05,
+      "loss": 0.0119,
+      "step": 9715
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7245762711864407e-05,
+      "loss": 0.2003,
+      "step": 9716
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7243408662900188e-05,
+      "loss": 0.0118,
+      "step": 9717
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7241054613935972e-05,
+      "loss": 0.0829,
+      "step": 9718
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7238700564971752e-05,
+      "loss": 0.1004,
+      "step": 9719
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7236346516007533e-05,
+      "loss": 0.0724,
+      "step": 9720
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7233992467043317e-05,
+      "loss": 0.1034,
+      "step": 9721
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7231638418079097e-05,
+      "loss": 0.178,
+      "step": 9722
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7229284369114878e-05,
+      "loss": 0.143,
+      "step": 9723
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7226930320150662e-05,
+      "loss": 0.2605,
+      "step": 9724
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7224576271186442e-05,
+      "loss": 0.0785,
+      "step": 9725
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7222222222222223e-05,
+      "loss": 0.0292,
+      "step": 9726
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7219868173258007e-05,
+      "loss": 0.0173,
+      "step": 9727
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7217514124293787e-05,
+      "loss": 0.0959,
+      "step": 9728
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7215160075329564e-05,
+      "loss": 0.1266,
+      "step": 9729
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7212806026365352e-05,
+      "loss": 0.0701,
+      "step": 9730
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.721045197740113e-05,
+      "loss": 0.2725,
+      "step": 9731
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7208097928436916e-05,
+      "loss": 0.0514,
+      "step": 9732
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7205743879472693e-05,
+      "loss": 0.0913,
+      "step": 9733
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7203389830508474e-05,
+      "loss": 0.0169,
+      "step": 9734
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.720103578154426e-05,
+      "loss": 0.0799,
+      "step": 9735
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.719868173258004e-05,
+      "loss": 0.0389,
+      "step": 9736
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.719632768361582e-05,
+      "loss": 0.1441,
+      "step": 9737
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7193973634651603e-05,
+      "loss": 0.0368,
+      "step": 9738
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7191619585687383e-05,
+      "loss": 0.056,
+      "step": 9739
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7189265536723164e-05,
+      "loss": 0.0342,
+      "step": 9740
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7186911487758948e-05,
+      "loss": 0.2147,
+      "step": 9741
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.718455743879473e-05,
+      "loss": 0.0959,
+      "step": 9742
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.718220338983051e-05,
+      "loss": 0.084,
+      "step": 9743
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7179849340866293e-05,
+      "loss": 0.0526,
+      "step": 9744
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7177495291902073e-05,
+      "loss": 0.1271,
+      "step": 9745
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7175141242937857e-05,
+      "loss": 0.0904,
+      "step": 9746
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7172787193973638e-05,
+      "loss": 0.1948,
+      "step": 9747
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7170433145009415e-05,
+      "loss": 0.1173,
+      "step": 9748
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7168079096045202e-05,
+      "loss": 0.206,
+      "step": 9749
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.716572504708098e-05,
+      "loss": 0.1003,
+      "step": 9750
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.716337099811676e-05,
+      "loss": 0.047,
+      "step": 9751
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7161016949152547e-05,
+      "loss": 0.1383,
+      "step": 9752
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7158662900188324e-05,
+      "loss": 0.1186,
+      "step": 9753
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7156308851224105e-05,
+      "loss": 0.0289,
+      "step": 9754
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.715395480225989e-05,
+      "loss": 0.1991,
+      "step": 9755
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.715160075329567e-05,
+      "loss": 0.1757,
+      "step": 9756
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.714924670433145e-05,
+      "loss": 0.1109,
+      "step": 9757
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7146892655367234e-05,
+      "loss": 0.0461,
+      "step": 9758
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7144538606403014e-05,
+      "loss": 0.0658,
+      "step": 9759
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7142184557438795e-05,
+      "loss": 0.1991,
+      "step": 9760
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.713983050847458e-05,
+      "loss": 0.0764,
+      "step": 9761
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.713747645951036e-05,
+      "loss": 0.0378,
+      "step": 9762
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7135122410546143e-05,
+      "loss": 0.0248,
+      "step": 9763
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7132768361581924e-05,
+      "loss": 0.0395,
+      "step": 9764
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.71304143126177e-05,
+      "loss": 0.3064,
+      "step": 9765
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.712806026365349e-05,
+      "loss": 0.0251,
+      "step": 9766
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7125706214689265e-05,
+      "loss": 0.1384,
+      "step": 9767
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7123352165725046e-05,
+      "loss": 0.0325,
+      "step": 9768
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7120998116760833e-05,
+      "loss": 0.1049,
+      "step": 9769
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.711864406779661e-05,
+      "loss": 0.0453,
+      "step": 9770
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.711629001883239e-05,
+      "loss": 0.0752,
+      "step": 9771
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7113935969868175e-05,
+      "loss": 0.1688,
+      "step": 9772
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7111581920903955e-05,
+      "loss": 0.0997,
+      "step": 9773
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7109227871939736e-05,
+      "loss": 0.023,
+      "step": 9774
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.710687382297552e-05,
+      "loss": 0.0805,
+      "step": 9775
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.71045197740113e-05,
+      "loss": 0.126,
+      "step": 9776
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.710216572504708e-05,
+      "loss": 0.1151,
+      "step": 9777
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7099811676082865e-05,
+      "loss": 0.092,
+      "step": 9778
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7097457627118645e-05,
+      "loss": 0.1113,
+      "step": 9779
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.709510357815443e-05,
+      "loss": 0.1035,
+      "step": 9780
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.709274952919021e-05,
+      "loss": 0.1298,
+      "step": 9781
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7090395480225987e-05,
+      "loss": 0.0629,
+      "step": 9782
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7088041431261774e-05,
+      "loss": 0.0734,
+      "step": 9783
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.708568738229755e-05,
+      "loss": 0.1638,
+      "step": 9784
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7083333333333332e-05,
+      "loss": 0.0535,
+      "step": 9785
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.708097928436912e-05,
+      "loss": 0.0822,
+      "step": 9786
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7078625235404897e-05,
+      "loss": 0.1198,
+      "step": 9787
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7076271186440677e-05,
+      "loss": 0.0783,
+      "step": 9788
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.707391713747646e-05,
+      "loss": 0.1297,
+      "step": 9789
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.707156308851224e-05,
+      "loss": 0.1167,
+      "step": 9790
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7069209039548022e-05,
+      "loss": 0.0457,
+      "step": 9791
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7066854990583806e-05,
+      "loss": 0.0621,
+      "step": 9792
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7064500941619587e-05,
+      "loss": 0.0551,
+      "step": 9793
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.706214689265537e-05,
+      "loss": 0.1872,
+      "step": 9794
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.705979284369115e-05,
+      "loss": 0.097,
+      "step": 9795
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.705743879472693e-05,
+      "loss": 0.1527,
+      "step": 9796
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7055084745762715e-05,
+      "loss": 0.0501,
+      "step": 9797
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7052730696798496e-05,
+      "loss": 0.0604,
+      "step": 9798
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7050376647834273e-05,
+      "loss": 0.0864,
+      "step": 9799
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.704802259887006e-05,
+      "loss": 0.2167,
+      "step": 9800
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.704566854990584e-05,
+      "loss": 0.149,
+      "step": 9801
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7043314500941618e-05,
+      "loss": 0.2203,
+      "step": 9802
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7040960451977405e-05,
+      "loss": 0.2931,
+      "step": 9803
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7038606403013183e-05,
+      "loss": 0.0749,
+      "step": 9804
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.7036252354048963e-05,
+      "loss": 0.1436,
+      "step": 9805
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.7033898305084747e-05,
+      "loss": 0.1697,
+      "step": 9806
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.7031544256120528e-05,
+      "loss": 0.1978,
+      "step": 9807
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.7029190207156308e-05,
+      "loss": 0.0354,
+      "step": 9808
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.7026836158192092e-05,
+      "loss": 0.0374,
+      "step": 9809
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.7024482109227873e-05,
+      "loss": 0.1749,
+      "step": 9810
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.7022128060263657e-05,
+      "loss": 0.2612,
+      "step": 9811
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.7019774011299437e-05,
+      "loss": 0.1363,
+      "step": 9812
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.7017419962335218e-05,
+      "loss": 0.1299,
+      "step": 9813
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.7015065913371e-05,
+      "loss": 0.1085,
+      "step": 9814
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.7012711864406782e-05,
+      "loss": 0.0788,
+      "step": 9815
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.701035781544256e-05,
+      "loss": 0.0468,
+      "step": 9816
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.7008003766478347e-05,
+      "loss": 0.026,
+      "step": 9817
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.7005649717514127e-05,
+      "loss": 0.0572,
+      "step": 9818
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.7003295668549904e-05,
+      "loss": 0.1503,
+      "step": 9819
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.700094161958569e-05,
+      "loss": 0.0851,
+      "step": 9820
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.699858757062147e-05,
+      "loss": 0.0545,
+      "step": 9821
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.699623352165725e-05,
+      "loss": 0.0905,
+      "step": 9822
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6993879472693033e-05,
+      "loss": 0.1242,
+      "step": 9823
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6991525423728814e-05,
+      "loss": 0.1086,
+      "step": 9824
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6989171374764598e-05,
+      "loss": 0.1352,
+      "step": 9825
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6986817325800378e-05,
+      "loss": 0.1524,
+      "step": 9826
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.698446327683616e-05,
+      "loss": 0.0414,
+      "step": 9827
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6982109227871943e-05,
+      "loss": 0.0375,
+      "step": 9828
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6979755178907723e-05,
+      "loss": 0.1408,
+      "step": 9829
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6977401129943504e-05,
+      "loss": 0.0649,
+      "step": 9830
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6975047080979288e-05,
+      "loss": 0.0203,
+      "step": 9831
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6972693032015068e-05,
+      "loss": 0.0405,
+      "step": 9832
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6970338983050845e-05,
+      "loss": 0.1521,
+      "step": 9833
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6967984934086633e-05,
+      "loss": 0.1368,
+      "step": 9834
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6965630885122413e-05,
+      "loss": 0.1881,
+      "step": 9835
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.696327683615819e-05,
+      "loss": 0.0192,
+      "step": 9836
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6960922787193978e-05,
+      "loss": 0.1684,
+      "step": 9837
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6958568738229755e-05,
+      "loss": 0.0572,
+      "step": 9838
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6956214689265535e-05,
+      "loss": 0.0434,
+      "step": 9839
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.695386064030132e-05,
+      "loss": 0.0891,
+      "step": 9840
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.69515065913371e-05,
+      "loss": 0.0844,
+      "step": 9841
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6949152542372884e-05,
+      "loss": 0.0947,
+      "step": 9842
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6946798493408664e-05,
+      "loss": 0.1599,
+      "step": 9843
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6944444444444445e-05,
+      "loss": 0.0358,
+      "step": 9844
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.694209039548023e-05,
+      "loss": 0.155,
+      "step": 9845
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.693973634651601e-05,
+      "loss": 0.2527,
+      "step": 9846
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.693738229755179e-05,
+      "loss": 0.0843,
+      "step": 9847
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6935028248587574e-05,
+      "loss": 0.1581,
+      "step": 9848
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6932674199623354e-05,
+      "loss": 0.0483,
+      "step": 9849
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.693032015065913e-05,
+      "loss": 0.0884,
+      "step": 9850
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.692796610169492e-05,
+      "loss": 0.0361,
+      "step": 9851
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.69256120527307e-05,
+      "loss": 0.0895,
+      "step": 9852
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6923258003766476e-05,
+      "loss": 0.0501,
+      "step": 9853
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6920903954802264e-05,
+      "loss": 0.1611,
+      "step": 9854
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.691854990583804e-05,
+      "loss": 0.1162,
+      "step": 9855
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.691619585687382e-05,
+      "loss": 0.1466,
+      "step": 9856
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6913841807909605e-05,
+      "loss": 0.2251,
+      "step": 9857
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6911487758945386e-05,
+      "loss": 0.0298,
+      "step": 9858
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.690913370998117e-05,
+      "loss": 0.1603,
+      "step": 9859
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.690677966101695e-05,
+      "loss": 0.0507,
+      "step": 9860
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.690442561205273e-05,
+      "loss": 0.0417,
+      "step": 9861
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6902071563088515e-05,
+      "loss": 0.0836,
+      "step": 9862
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6899717514124295e-05,
+      "loss": 0.0815,
+      "step": 9863
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6897363465160076e-05,
+      "loss": 0.0784,
+      "step": 9864
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.689500941619586e-05,
+      "loss": 0.1019,
+      "step": 9865
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.689265536723164e-05,
+      "loss": 0.1332,
+      "step": 9866
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6890301318267417e-05,
+      "loss": 0.1841,
+      "step": 9867
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6887947269303205e-05,
+      "loss": 0.1685,
+      "step": 9868
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6885593220338985e-05,
+      "loss": 0.0828,
+      "step": 9869
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6883239171374762e-05,
+      "loss": 0.0404,
+      "step": 9870
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.688088512241055e-05,
+      "loss": 0.0622,
+      "step": 9871
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6878531073446327e-05,
+      "loss": 0.0283,
+      "step": 9872
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6876177024482114e-05,
+      "loss": 0.1411,
+      "step": 9873
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.687382297551789e-05,
+      "loss": 0.0989,
+      "step": 9874
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6871468926553672e-05,
+      "loss": 0.0403,
+      "step": 9875
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6869114877589456e-05,
+      "loss": 0.2459,
+      "step": 9876
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6866760828625236e-05,
+      "loss": 0.0935,
+      "step": 9877
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6864406779661017e-05,
+      "loss": 0.1036,
+      "step": 9878
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.68620527306968e-05,
+      "loss": 0.1454,
+      "step": 9879
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.685969868173258e-05,
+      "loss": 0.184,
+      "step": 9880
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6857344632768362e-05,
+      "loss": 0.0542,
+      "step": 9881
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6854990583804146e-05,
+      "loss": 0.1026,
+      "step": 9882
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6852636534839926e-05,
+      "loss": 0.0613,
+      "step": 9883
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6850282485875707e-05,
+      "loss": 0.073,
+      "step": 9884
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.684792843691149e-05,
+      "loss": 0.2888,
+      "step": 9885
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.684557438794727e-05,
+      "loss": 0.1239,
+      "step": 9886
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.684322033898305e-05,
+      "loss": 0.1091,
+      "step": 9887
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6840866290018836e-05,
+      "loss": 0.0531,
+      "step": 9888
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6838512241054613e-05,
+      "loss": 0.1318,
+      "step": 9889
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.68361581920904e-05,
+      "loss": 0.1356,
+      "step": 9890
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6833804143126177e-05,
+      "loss": 0.1247,
+      "step": 9891
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6831450094161958e-05,
+      "loss": 0.0946,
+      "step": 9892
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6829096045197745e-05,
+      "loss": 0.0892,
+      "step": 9893
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6826741996233522e-05,
+      "loss": 0.1759,
+      "step": 9894
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6824387947269303e-05,
+      "loss": 0.0432,
+      "step": 9895
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6822033898305087e-05,
+      "loss": 0.0207,
+      "step": 9896
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6819679849340867e-05,
+      "loss": 0.0827,
+      "step": 9897
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6817325800376648e-05,
+      "loss": 0.0789,
+      "step": 9898
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6814971751412432e-05,
+      "loss": 0.108,
+      "step": 9899
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6812617702448212e-05,
+      "loss": 0.2794,
+      "step": 9900
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6810263653483993e-05,
+      "loss": 0.1417,
+      "step": 9901
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6807909604519777e-05,
+      "loss": 0.0637,
+      "step": 9902
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6805555555555557e-05,
+      "loss": 0.0335,
+      "step": 9903
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.680320150659134e-05,
+      "loss": 0.0605,
+      "step": 9904
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6800847457627122e-05,
+      "loss": 0.14,
+      "step": 9905
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.67984934086629e-05,
+      "loss": 0.0889,
+      "step": 9906
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6796139359698686e-05,
+      "loss": 0.0889,
+      "step": 9907
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6793785310734463e-05,
+      "loss": 0.0825,
+      "step": 9908
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6791431261770244e-05,
+      "loss": 0.1055,
+      "step": 9909
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.678907721280603e-05,
+      "loss": 0.0942,
+      "step": 9910
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.678672316384181e-05,
+      "loss": 0.0478,
+      "step": 9911
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.678436911487759e-05,
+      "loss": 0.1629,
+      "step": 9912
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6782015065913373e-05,
+      "loss": 0.0303,
+      "step": 9913
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6779661016949153e-05,
+      "loss": 0.0565,
+      "step": 9914
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6777306967984934e-05,
+      "loss": 0.0671,
+      "step": 9915
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6774952919020718e-05,
+      "loss": 0.04,
+      "step": 9916
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.67725988700565e-05,
+      "loss": 0.1877,
+      "step": 9917
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.677024482109228e-05,
+      "loss": 0.0604,
+      "step": 9918
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6767890772128063e-05,
+      "loss": 0.0887,
+      "step": 9919
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6765536723163843e-05,
+      "loss": 0.1395,
+      "step": 9920
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6763182674199627e-05,
+      "loss": 0.0718,
+      "step": 9921
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6760828625235408e-05,
+      "loss": 0.0707,
+      "step": 9922
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6758474576271185e-05,
+      "loss": 0.1355,
+      "step": 9923
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6756120527306972e-05,
+      "loss": 0.0748,
+      "step": 9924
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.675376647834275e-05,
+      "loss": 0.0265,
+      "step": 9925
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.675141242937853e-05,
+      "loss": 0.1307,
+      "step": 9926
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6749058380414317e-05,
+      "loss": 0.0806,
+      "step": 9927
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6746704331450094e-05,
+      "loss": 0.0903,
+      "step": 9928
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6744350282485875e-05,
+      "loss": 0.132,
+      "step": 9929
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.674199623352166e-05,
+      "loss": 0.0293,
+      "step": 9930
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.673964218455744e-05,
+      "loss": 0.1042,
+      "step": 9931
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.673728813559322e-05,
+      "loss": 0.0226,
+      "step": 9932
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6734934086629004e-05,
+      "loss": 0.181,
+      "step": 9933
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6732580037664784e-05,
+      "loss": 0.2437,
+      "step": 9934
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6730225988700565e-05,
+      "loss": 0.2786,
+      "step": 9935
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.672787193973635e-05,
+      "loss": 0.1043,
+      "step": 9936
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.672551789077213e-05,
+      "loss": 0.1276,
+      "step": 9937
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6723163841807913e-05,
+      "loss": 0.1475,
+      "step": 9938
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6720809792843694e-05,
+      "loss": 0.0913,
+      "step": 9939
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.671845574387947e-05,
+      "loss": 0.1104,
+      "step": 9940
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.671610169491526e-05,
+      "loss": 0.0694,
+      "step": 9941
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6713747645951036e-05,
+      "loss": 0.1501,
+      "step": 9942
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6711393596986816e-05,
+      "loss": 0.1249,
+      "step": 9943
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6709039548022603e-05,
+      "loss": 0.0746,
+      "step": 9944
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.670668549905838e-05,
+      "loss": 0.0325,
+      "step": 9945
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.670433145009416e-05,
+      "loss": 0.0473,
+      "step": 9946
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6701977401129945e-05,
+      "loss": 0.1117,
+      "step": 9947
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6699623352165726e-05,
+      "loss": 0.0716,
+      "step": 9948
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6697269303201506e-05,
+      "loss": 0.0809,
+      "step": 9949
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.669491525423729e-05,
+      "loss": 0.0193,
+      "step": 9950
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.669256120527307e-05,
+      "loss": 0.0986,
+      "step": 9951
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6690207156308854e-05,
+      "loss": 0.0997,
+      "step": 9952
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6687853107344635e-05,
+      "loss": 0.066,
+      "step": 9953
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6685499058380416e-05,
+      "loss": 0.0396,
+      "step": 9954
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.66831450094162e-05,
+      "loss": 0.1225,
+      "step": 9955
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.668079096045198e-05,
+      "loss": 0.0166,
+      "step": 9956
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6678436911487757e-05,
+      "loss": 0.0645,
+      "step": 9957
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6676082862523544e-05,
+      "loss": 0.0425,
+      "step": 9958
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6673728813559325e-05,
+      "loss": 0.0575,
+      "step": 9959
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6671374764595102e-05,
+      "loss": 0.0591,
+      "step": 9960
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.666902071563089e-05,
+      "loss": 0.1173,
+      "step": 9961
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.1712,
+      "step": 9962
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6664312617702447e-05,
+      "loss": 0.1694,
+      "step": 9963
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.666195856873823e-05,
+      "loss": 0.1486,
+      "step": 9964
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.665960451977401e-05,
+      "loss": 0.0953,
+      "step": 9965
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6657250470809792e-05,
+      "loss": 0.0925,
+      "step": 9966
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6654896421845576e-05,
+      "loss": 0.1309,
+      "step": 9967
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6652542372881357e-05,
+      "loss": 0.2784,
+      "step": 9968
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.665018832391714e-05,
+      "loss": 0.1381,
+      "step": 9969
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.664783427495292e-05,
+      "loss": 0.1059,
+      "step": 9970
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.66454802259887e-05,
+      "loss": 0.05,
+      "step": 9971
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6643126177024486e-05,
+      "loss": 0.0778,
+      "step": 9972
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6640772128060266e-05,
+      "loss": 0.0369,
+      "step": 9973
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6638418079096043e-05,
+      "loss": 0.1175,
+      "step": 9974
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.663606403013183e-05,
+      "loss": 0.0546,
+      "step": 9975
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.663370998116761e-05,
+      "loss": 0.0649,
+      "step": 9976
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6631355932203388e-05,
+      "loss": 0.0551,
+      "step": 9977
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6629001883239175e-05,
+      "loss": 0.1175,
+      "step": 9978
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6626647834274953e-05,
+      "loss": 0.1829,
+      "step": 9979
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6624293785310733e-05,
+      "loss": 0.1829,
+      "step": 9980
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6621939736346517e-05,
+      "loss": 0.19,
+      "step": 9981
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6619585687382298e-05,
+      "loss": 0.0418,
+      "step": 9982
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.661723163841808e-05,
+      "loss": 0.0962,
+      "step": 9983
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6614877589453862e-05,
+      "loss": 0.1413,
+      "step": 9984
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6612523540489643e-05,
+      "loss": 0.1322,
+      "step": 9985
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6610169491525427e-05,
+      "loss": 0.0387,
+      "step": 9986
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6607815442561207e-05,
+      "loss": 0.1462,
+      "step": 9987
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6605461393596988e-05,
+      "loss": 0.1327,
+      "step": 9988
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.660310734463277e-05,
+      "loss": 0.0116,
+      "step": 9989
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6600753295668552e-05,
+      "loss": 0.0509,
+      "step": 9990
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.659839924670433e-05,
+      "loss": 0.1085,
+      "step": 9991
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6596045197740117e-05,
+      "loss": 0.1311,
+      "step": 9992
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6593691148775897e-05,
+      "loss": 0.2874,
+      "step": 9993
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6591337099811674e-05,
+      "loss": 0.1227,
+      "step": 9994
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.658898305084746e-05,
+      "loss": 0.0646,
+      "step": 9995
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.658662900188324e-05,
+      "loss": 0.154,
+      "step": 9996
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.658427495291902e-05,
+      "loss": 0.1847,
+      "step": 9997
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6581920903954803e-05,
+      "loss": 0.1534,
+      "step": 9998
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6579566854990584e-05,
+      "loss": 0.163,
+      "step": 9999
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6577212806026368e-05,
+      "loss": 0.0703,
+      "step": 10000
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6574858757062148e-05,
+      "loss": 0.0883,
+      "step": 10001
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.657250470809793e-05,
+      "loss": 0.0902,
+      "step": 10002
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6570150659133713e-05,
+      "loss": 0.0589,
+      "step": 10003
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6567796610169493e-05,
+      "loss": 0.1151,
+      "step": 10004
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6565442561205274e-05,
+      "loss": 0.0863,
+      "step": 10005
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6563088512241058e-05,
+      "loss": 0.0309,
+      "step": 10006
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6560734463276838e-05,
+      "loss": 0.1138,
+      "step": 10007
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6558380414312615e-05,
+      "loss": 0.0106,
+      "step": 10008
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6556026365348403e-05,
+      "loss": 0.0896,
+      "step": 10009
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6553672316384183e-05,
+      "loss": 0.0634,
+      "step": 10010
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.655131826741996e-05,
+      "loss": 0.0458,
+      "step": 10011
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6548964218455748e-05,
+      "loss": 0.0154,
+      "step": 10012
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6546610169491525e-05,
+      "loss": 0.1347,
+      "step": 10013
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6544256120527305e-05,
+      "loss": 0.0673,
+      "step": 10014
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.654190207156309e-05,
+      "loss": 0.0915,
+      "step": 10015
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.653954802259887e-05,
+      "loss": 0.1814,
+      "step": 10016
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6537193973634654e-05,
+      "loss": 0.1635,
+      "step": 10017
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6534839924670434e-05,
+      "loss": 0.0264,
+      "step": 10018
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6532485875706215e-05,
+      "loss": 0.1766,
+      "step": 10019
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6530131826742e-05,
+      "loss": 0.1757,
+      "step": 10020
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.652777777777778e-05,
+      "loss": 0.0657,
+      "step": 10021
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.652542372881356e-05,
+      "loss": 0.027,
+      "step": 10022
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6523069679849344e-05,
+      "loss": 0.027,
+      "step": 10023
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6520715630885124e-05,
+      "loss": 0.0318,
+      "step": 10024
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.65183615819209e-05,
+      "loss": 0.0531,
+      "step": 10025
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.651600753295669e-05,
+      "loss": 0.0838,
+      "step": 10026
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.651365348399247e-05,
+      "loss": 0.1014,
+      "step": 10027
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6511299435028246e-05,
+      "loss": 0.1359,
+      "step": 10028
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6508945386064034e-05,
+      "loss": 0.1373,
+      "step": 10029
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.650659133709981e-05,
+      "loss": 0.1117,
+      "step": 10030
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6504237288135598e-05,
+      "loss": 0.0118,
+      "step": 10031
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6501883239171375e-05,
+      "loss": 0.072,
+      "step": 10032
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6499529190207156e-05,
+      "loss": 0.0508,
+      "step": 10033
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.649717514124294e-05,
+      "loss": 0.0449,
+      "step": 10034
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.649482109227872e-05,
+      "loss": 0.0626,
+      "step": 10035
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.64924670433145e-05,
+      "loss": 0.0255,
+      "step": 10036
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6490112994350285e-05,
+      "loss": 0.0652,
+      "step": 10037
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6487758945386065e-05,
+      "loss": 0.1605,
+      "step": 10038
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6485404896421846e-05,
+      "loss": 0.0192,
+      "step": 10039
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.648305084745763e-05,
+      "loss": 0.2082,
+      "step": 10040
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.648069679849341e-05,
+      "loss": 0.0815,
+      "step": 10041
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.647834274952919e-05,
+      "loss": 0.0244,
+      "step": 10042
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6475988700564975e-05,
+      "loss": 0.1448,
+      "step": 10043
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6473634651600755e-05,
+      "loss": 0.1048,
+      "step": 10044
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6471280602636532e-05,
+      "loss": 0.0914,
+      "step": 10045
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.646892655367232e-05,
+      "loss": 0.0457,
+      "step": 10046
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6466572504708097e-05,
+      "loss": 0.0593,
+      "step": 10047
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6464218455743884e-05,
+      "loss": 0.1096,
+      "step": 10048
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.646186440677966e-05,
+      "loss": 0.1536,
+      "step": 10049
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6459510357815442e-05,
+      "loss": 0.1219,
+      "step": 10050
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.645715630885123e-05,
+      "loss": 0.2012,
+      "step": 10051
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6454802259887006e-05,
+      "loss": 0.1468,
+      "step": 10052
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6452448210922787e-05,
+      "loss": 0.0955,
+      "step": 10053
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.645009416195857e-05,
+      "loss": 0.106,
+      "step": 10054
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.644774011299435e-05,
+      "loss": 0.0847,
+      "step": 10055
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6445386064030132e-05,
+      "loss": 0.1548,
+      "step": 10056
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6443032015065916e-05,
+      "loss": 0.0445,
+      "step": 10057
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6440677966101696e-05,
+      "loss": 0.0653,
+      "step": 10058
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6438323917137477e-05,
+      "loss": 0.0418,
+      "step": 10059
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.643596986817326e-05,
+      "loss": 0.0736,
+      "step": 10060
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.643361581920904e-05,
+      "loss": 0.1246,
+      "step": 10061
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6431261770244825e-05,
+      "loss": 0.0393,
+      "step": 10062
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6428907721280606e-05,
+      "loss": 0.0682,
+      "step": 10063
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6426553672316383e-05,
+      "loss": 0.1211,
+      "step": 10064
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.642419962335217e-05,
+      "loss": 0.0528,
+      "step": 10065
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6421845574387947e-05,
+      "loss": 0.0679,
+      "step": 10066
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6419491525423728e-05,
+      "loss": 0.0606,
+      "step": 10067
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6417137476459515e-05,
+      "loss": 0.1332,
+      "step": 10068
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6414783427495292e-05,
+      "loss": 0.0189,
+      "step": 10069
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6412429378531073e-05,
+      "loss": 0.0271,
+      "step": 10070
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6410075329566857e-05,
+      "loss": 0.1008,
+      "step": 10071
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6407721280602637e-05,
+      "loss": 0.0535,
+      "step": 10072
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6405367231638418e-05,
+      "loss": 0.0989,
+      "step": 10073
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6403013182674202e-05,
+      "loss": 0.0208,
+      "step": 10074
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6400659133709982e-05,
+      "loss": 0.1753,
+      "step": 10075
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6398305084745763e-05,
+      "loss": 0.0133,
+      "step": 10076
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6395951035781547e-05,
+      "loss": 0.1355,
+      "step": 10077
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6393596986817327e-05,
+      "loss": 0.0866,
+      "step": 10078
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.639124293785311e-05,
+      "loss": 0.0994,
+      "step": 10079
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6388888888888892e-05,
+      "loss": 0.0861,
+      "step": 10080
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.638653483992467e-05,
+      "loss": 0.0925,
+      "step": 10081
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6384180790960456e-05,
+      "loss": 0.0368,
+      "step": 10082
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6381826741996233e-05,
+      "loss": 0.2743,
+      "step": 10083
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6379472693032014e-05,
+      "loss": 0.1162,
+      "step": 10084
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.63771186440678e-05,
+      "loss": 0.0306,
+      "step": 10085
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.637476459510358e-05,
+      "loss": 0.1131,
+      "step": 10086
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.637241054613936e-05,
+      "loss": 0.1099,
+      "step": 10087
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6370056497175143e-05,
+      "loss": 0.1047,
+      "step": 10088
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6367702448210923e-05,
+      "loss": 0.32,
+      "step": 10089
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6365348399246704e-05,
+      "loss": 0.1314,
+      "step": 10090
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6362994350282488e-05,
+      "loss": 0.0993,
+      "step": 10091
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.636064030131827e-05,
+      "loss": 0.2125,
+      "step": 10092
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.635828625235405e-05,
+      "loss": 0.072,
+      "step": 10093
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6355932203389833e-05,
+      "loss": 0.0831,
+      "step": 10094
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6353578154425613e-05,
+      "loss": 0.0434,
+      "step": 10095
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6351224105461397e-05,
+      "loss": 0.037,
+      "step": 10096
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6348870056497178e-05,
+      "loss": 0.1681,
+      "step": 10097
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6346516007532955e-05,
+      "loss": 0.0659,
+      "step": 10098
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6344161958568742e-05,
+      "loss": 0.137,
+      "step": 10099
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.634180790960452e-05,
+      "loss": 0.1228,
+      "step": 10100
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.63394538606403e-05,
+      "loss": 0.0868,
+      "step": 10101
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6337099811676087e-05,
+      "loss": 0.1575,
+      "step": 10102
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6334745762711865e-05,
+      "loss": 0.0353,
+      "step": 10103
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6332391713747645e-05,
+      "loss": 0.0589,
+      "step": 10104
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.633003766478343e-05,
+      "loss": 0.0504,
+      "step": 10105
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.632768361581921e-05,
+      "loss": 0.1437,
+      "step": 10106
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.632532956685499e-05,
+      "loss": 0.079,
+      "step": 10107
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6322975517890774e-05,
+      "loss": 0.0259,
+      "step": 10108
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6320621468926555e-05,
+      "loss": 0.1749,
+      "step": 10109
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.631826741996234e-05,
+      "loss": 0.0522,
+      "step": 10110
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.631591337099812e-05,
+      "loss": 0.0411,
+      "step": 10111
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.63135593220339e-05,
+      "loss": 0.0222,
+      "step": 10112
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6311205273069683e-05,
+      "loss": 0.1251,
+      "step": 10113
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6308851224105464e-05,
+      "loss": 0.0376,
+      "step": 10114
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.630649717514124e-05,
+      "loss": 0.0337,
+      "step": 10115
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.630414312617703e-05,
+      "loss": 0.0555,
+      "step": 10116
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.630178907721281e-05,
+      "loss": 0.1418,
+      "step": 10117
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6299435028248586e-05,
+      "loss": 0.0776,
+      "step": 10118
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6297080979284373e-05,
+      "loss": 0.0536,
+      "step": 10119
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.629472693032015e-05,
+      "loss": 0.0686,
+      "step": 10120
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.629237288135593e-05,
+      "loss": 0.089,
+      "step": 10121
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6290018832391715e-05,
+      "loss": 0.0512,
+      "step": 10122
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6287664783427496e-05,
+      "loss": 0.0334,
+      "step": 10123
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6285310734463276e-05,
+      "loss": 0.0587,
+      "step": 10124
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.628295668549906e-05,
+      "loss": 0.0428,
+      "step": 10125
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.628060263653484e-05,
+      "loss": 0.0199,
+      "step": 10126
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6278248587570624e-05,
+      "loss": 0.0606,
+      "step": 10127
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6275894538606405e-05,
+      "loss": 0.0633,
+      "step": 10128
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6273540489642186e-05,
+      "loss": 0.0337,
+      "step": 10129
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.627118644067797e-05,
+      "loss": 0.0946,
+      "step": 10130
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.626883239171375e-05,
+      "loss": 0.1864,
+      "step": 10131
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6266478342749527e-05,
+      "loss": 0.0265,
+      "step": 10132
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6264124293785314e-05,
+      "loss": 0.0746,
+      "step": 10133
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6261770244821095e-05,
+      "loss": 0.1289,
+      "step": 10134
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6259416195856872e-05,
+      "loss": 0.0895,
+      "step": 10135
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.625706214689266e-05,
+      "loss": 0.2281,
+      "step": 10136
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6254708097928437e-05,
+      "loss": 0.034,
+      "step": 10137
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6252354048964217e-05,
+      "loss": 0.0668,
+      "step": 10138
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.625e-05,
+      "loss": 0.1092,
+      "step": 10139
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.624764595103578e-05,
+      "loss": 0.0957,
+      "step": 10140
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6245291902071566e-05,
+      "loss": 0.1286,
+      "step": 10141
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6242937853107346e-05,
+      "loss": 0.0814,
+      "step": 10142
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6240583804143127e-05,
+      "loss": 0.2286,
+      "step": 10143
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.623822975517891e-05,
+      "loss": 0.0354,
+      "step": 10144
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.623587570621469e-05,
+      "loss": 0.0837,
+      "step": 10145
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.623352165725047e-05,
+      "loss": 0.056,
+      "step": 10146
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6231167608286256e-05,
+      "loss": 0.196,
+      "step": 10147
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6228813559322036e-05,
+      "loss": 0.0957,
+      "step": 10148
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6226459510357813e-05,
+      "loss": 0.1118,
+      "step": 10149
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.62241054613936e-05,
+      "loss": 0.0523,
+      "step": 10150
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.622175141242938e-05,
+      "loss": 0.1069,
+      "step": 10151
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6219397363465158e-05,
+      "loss": 0.0486,
+      "step": 10152
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6217043314500946e-05,
+      "loss": 0.1385,
+      "step": 10153
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6214689265536723e-05,
+      "loss": 0.0528,
+      "step": 10154
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6212335216572503e-05,
+      "loss": 0.0919,
+      "step": 10155
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6209981167608287e-05,
+      "loss": 0.0954,
+      "step": 10156
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6207627118644068e-05,
+      "loss": 0.055,
+      "step": 10157
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.620527306967985e-05,
+      "loss": 0.0935,
+      "step": 10158
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6202919020715632e-05,
+      "loss": 0.1081,
+      "step": 10159
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6200564971751413e-05,
+      "loss": 0.0592,
+      "step": 10160
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6198210922787197e-05,
+      "loss": 0.1785,
+      "step": 10161
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6195856873822977e-05,
+      "loss": 0.2021,
+      "step": 10162
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6193502824858758e-05,
+      "loss": 0.0501,
+      "step": 10163
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.619114877589454e-05,
+      "loss": 0.0629,
+      "step": 10164
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6188794726930322e-05,
+      "loss": 0.0328,
+      "step": 10165
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.61864406779661e-05,
+      "loss": 0.1181,
+      "step": 10166
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6184086629001887e-05,
+      "loss": 0.1163,
+      "step": 10167
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6181732580037667e-05,
+      "loss": 0.1277,
+      "step": 10168
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6179378531073444e-05,
+      "loss": 0.1512,
+      "step": 10169
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.617702448210923e-05,
+      "loss": 0.013,
+      "step": 10170
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.617467043314501e-05,
+      "loss": 0.0276,
+      "step": 10171
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.617231638418079e-05,
+      "loss": 0.0826,
+      "step": 10172
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6169962335216573e-05,
+      "loss": 0.1484,
+      "step": 10173
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6167608286252354e-05,
+      "loss": 0.0682,
+      "step": 10174
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6165254237288138e-05,
+      "loss": 0.0401,
+      "step": 10175
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6162900188323918e-05,
+      "loss": 0.0416,
+      "step": 10176
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.61605461393597e-05,
+      "loss": 0.1709,
+      "step": 10177
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6158192090395483e-05,
+      "loss": 0.0681,
+      "step": 10178
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6155838041431263e-05,
+      "loss": 0.0132,
+      "step": 10179
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6153483992467044e-05,
+      "loss": 0.075,
+      "step": 10180
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6151129943502828e-05,
+      "loss": 0.1411,
+      "step": 10181
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6148775894538608e-05,
+      "loss": 0.0874,
+      "step": 10182
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6146421845574385e-05,
+      "loss": 0.0254,
+      "step": 10183
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6144067796610173e-05,
+      "loss": 0.197,
+      "step": 10184
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6141713747645953e-05,
+      "loss": 0.0933,
+      "step": 10185
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.613935969868173e-05,
+      "loss": 0.0547,
+      "step": 10186
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6137005649717518e-05,
+      "loss": 0.1343,
+      "step": 10187
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6134651600753295e-05,
+      "loss": 0.0307,
+      "step": 10188
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6132297551789082e-05,
+      "loss": 0.113,
+      "step": 10189
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.612994350282486e-05,
+      "loss": 0.0557,
+      "step": 10190
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.612758945386064e-05,
+      "loss": 0.1383,
+      "step": 10191
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6125235404896424e-05,
+      "loss": 0.1055,
+      "step": 10192
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6122881355932204e-05,
+      "loss": 0.0585,
+      "step": 10193
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6120527306967985e-05,
+      "loss": 0.106,
+      "step": 10194
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.611817325800377e-05,
+      "loss": 0.043,
+      "step": 10195
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.611581920903955e-05,
+      "loss": 0.1101,
+      "step": 10196
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.611346516007533e-05,
+      "loss": 0.0578,
+      "step": 10197
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6111111111111114e-05,
+      "loss": 0.0394,
+      "step": 10198
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6108757062146894e-05,
+      "loss": 0.0335,
+      "step": 10199
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6106403013182675e-05,
+      "loss": 0.1258,
+      "step": 10200
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.610404896421846e-05,
+      "loss": 0.0194,
+      "step": 10201
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.610169491525424e-05,
+      "loss": 0.1211,
+      "step": 10202
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6099340866290016e-05,
+      "loss": 0.1268,
+      "step": 10203
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6096986817325804e-05,
+      "loss": 0.1506,
+      "step": 10204
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.609463276836158e-05,
+      "loss": 0.2054,
+      "step": 10205
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6092278719397368e-05,
+      "loss": 0.1482,
+      "step": 10206
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6089924670433145e-05,
+      "loss": 0.1745,
+      "step": 10207
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6087570621468926e-05,
+      "loss": 0.1077,
+      "step": 10208
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6085216572504713e-05,
+      "loss": 0.0179,
+      "step": 10209
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.608286252354049e-05,
+      "loss": 0.0771,
+      "step": 10210
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.608050847457627e-05,
+      "loss": 0.1005,
+      "step": 10211
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6078154425612055e-05,
+      "loss": 0.0391,
+      "step": 10212
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6075800376647835e-05,
+      "loss": 0.0839,
+      "step": 10213
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6073446327683616e-05,
+      "loss": 0.1458,
+      "step": 10214
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.60710922787194e-05,
+      "loss": 0.1292,
+      "step": 10215
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.606873822975518e-05,
+      "loss": 0.135,
+      "step": 10216
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.606638418079096e-05,
+      "loss": 0.1608,
+      "step": 10217
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6064030131826745e-05,
+      "loss": 0.0423,
+      "step": 10218
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6061676082862525e-05,
+      "loss": 0.07,
+      "step": 10219
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.605932203389831e-05,
+      "loss": 0.0399,
+      "step": 10220
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.605696798493409e-05,
+      "loss": 0.0931,
+      "step": 10221
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6054613935969867e-05,
+      "loss": 0.07,
+      "step": 10222
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6052259887005654e-05,
+      "loss": 0.1534,
+      "step": 10223
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.604990583804143e-05,
+      "loss": 0.0482,
+      "step": 10224
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6047551789077212e-05,
+      "loss": 0.0778,
+      "step": 10225
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6045197740113e-05,
+      "loss": 0.049,
+      "step": 10226
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6042843691148776e-05,
+      "loss": 0.0779,
+      "step": 10227
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6040489642184557e-05,
+      "loss": 0.0953,
+      "step": 10228
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.603813559322034e-05,
+      "loss": 0.1108,
+      "step": 10229
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.603578154425612e-05,
+      "loss": 0.1029,
+      "step": 10230
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.6033427495291902e-05,
+      "loss": 0.0722,
+      "step": 10231
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.6031073446327686e-05,
+      "loss": 0.0293,
+      "step": 10232
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.6028719397363466e-05,
+      "loss": 0.0484,
+      "step": 10233
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.6026365348399247e-05,
+      "loss": 0.09,
+      "step": 10234
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.602401129943503e-05,
+      "loss": 0.1145,
+      "step": 10235
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.602165725047081e-05,
+      "loss": 0.1787,
+      "step": 10236
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.6019303201506595e-05,
+      "loss": 0.1271,
+      "step": 10237
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.6016949152542376e-05,
+      "loss": 0.0896,
+      "step": 10238
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.6014595103578153e-05,
+      "loss": 0.0226,
+      "step": 10239
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.601224105461394e-05,
+      "loss": 0.2563,
+      "step": 10240
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.6009887005649717e-05,
+      "loss": 0.0645,
+      "step": 10241
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.6007532956685498e-05,
+      "loss": 0.1622,
+      "step": 10242
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.6005178907721285e-05,
+      "loss": 0.0575,
+      "step": 10243
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.6002824858757062e-05,
+      "loss": 0.1309,
+      "step": 10244
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.6000470809792843e-05,
+      "loss": 0.0421,
+      "step": 10245
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5998116760828627e-05,
+      "loss": 0.0805,
+      "step": 10246
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5995762711864407e-05,
+      "loss": 0.092,
+      "step": 10247
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5993408662900188e-05,
+      "loss": 0.0754,
+      "step": 10248
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5991054613935972e-05,
+      "loss": 0.1331,
+      "step": 10249
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5988700564971752e-05,
+      "loss": 0.0867,
+      "step": 10250
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5986346516007533e-05,
+      "loss": 0.0951,
+      "step": 10251
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5983992467043317e-05,
+      "loss": 0.1796,
+      "step": 10252
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5981638418079097e-05,
+      "loss": 0.133,
+      "step": 10253
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.597928436911488e-05,
+      "loss": 0.0351,
+      "step": 10254
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5976930320150662e-05,
+      "loss": 0.077,
+      "step": 10255
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.597457627118644e-05,
+      "loss": 0.0506,
+      "step": 10256
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5972222222222226e-05,
+      "loss": 0.1347,
+      "step": 10257
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5969868173258003e-05,
+      "loss": 0.1188,
+      "step": 10258
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5967514124293784e-05,
+      "loss": 0.1193,
+      "step": 10259
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.596516007532957e-05,
+      "loss": 0.061,
+      "step": 10260
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.596280602636535e-05,
+      "loss": 0.0993,
+      "step": 10261
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.596045197740113e-05,
+      "loss": 0.0363,
+      "step": 10262
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5958097928436913e-05,
+      "loss": 0.0754,
+      "step": 10263
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5955743879472693e-05,
+      "loss": 0.1588,
+      "step": 10264
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5953389830508474e-05,
+      "loss": 0.0664,
+      "step": 10265
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5951035781544258e-05,
+      "loss": 0.1297,
+      "step": 10266
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.594868173258004e-05,
+      "loss": 0.1026,
+      "step": 10267
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5946327683615822e-05,
+      "loss": 0.1207,
+      "step": 10268
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5943973634651603e-05,
+      "loss": 0.0542,
+      "step": 10269
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5941619585687383e-05,
+      "loss": 0.0404,
+      "step": 10270
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5939265536723167e-05,
+      "loss": 0.1411,
+      "step": 10271
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5936911487758948e-05,
+      "loss": 0.0371,
+      "step": 10272
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5934557438794725e-05,
+      "loss": 0.0812,
+      "step": 10273
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5932203389830512e-05,
+      "loss": 0.0724,
+      "step": 10274
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5929849340866293e-05,
+      "loss": 0.0423,
+      "step": 10275
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.592749529190207e-05,
+      "loss": 0.0599,
+      "step": 10276
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5925141242937857e-05,
+      "loss": 0.0701,
+      "step": 10277
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5922787193973635e-05,
+      "loss": 0.153,
+      "step": 10278
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5920433145009415e-05,
+      "loss": 0.0992,
+      "step": 10279
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.59180790960452e-05,
+      "loss": 0.0959,
+      "step": 10280
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.591572504708098e-05,
+      "loss": 0.0911,
+      "step": 10281
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.591337099811676e-05,
+      "loss": 0.0398,
+      "step": 10282
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5911016949152544e-05,
+      "loss": 0.0419,
+      "step": 10283
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5908662900188325e-05,
+      "loss": 0.0295,
+      "step": 10284
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.590630885122411e-05,
+      "loss": 0.0673,
+      "step": 10285
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.590395480225989e-05,
+      "loss": 0.1994,
+      "step": 10286
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.590160075329567e-05,
+      "loss": 0.1251,
+      "step": 10287
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5899246704331453e-05,
+      "loss": 0.0451,
+      "step": 10288
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5896892655367234e-05,
+      "loss": 0.2198,
+      "step": 10289
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.589453860640301e-05,
+      "loss": 0.0273,
+      "step": 10290
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.58921845574388e-05,
+      "loss": 0.184,
+      "step": 10291
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.588983050847458e-05,
+      "loss": 0.2026,
+      "step": 10292
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5887476459510356e-05,
+      "loss": 0.1451,
+      "step": 10293
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5885122410546143e-05,
+      "loss": 0.0698,
+      "step": 10294
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.588276836158192e-05,
+      "loss": 0.0779,
+      "step": 10295
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.58804143126177e-05,
+      "loss": 0.2057,
+      "step": 10296
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5878060263653485e-05,
+      "loss": 0.0728,
+      "step": 10297
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5875706214689266e-05,
+      "loss": 0.1801,
+      "step": 10298
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.587335216572505e-05,
+      "loss": 0.0334,
+      "step": 10299
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.587099811676083e-05,
+      "loss": 0.0424,
+      "step": 10300
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.586864406779661e-05,
+      "loss": 0.0249,
+      "step": 10301
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5866290018832395e-05,
+      "loss": 0.1155,
+      "step": 10302
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5863935969868175e-05,
+      "loss": 0.176,
+      "step": 10303
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5861581920903956e-05,
+      "loss": 0.0127,
+      "step": 10304
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.585922787193974e-05,
+      "loss": 0.1963,
+      "step": 10305
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.585687382297552e-05,
+      "loss": 0.142,
+      "step": 10306
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5854519774011297e-05,
+      "loss": 0.1838,
+      "step": 10307
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5852165725047085e-05,
+      "loss": 0.2376,
+      "step": 10308
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5849811676082865e-05,
+      "loss": 0.0385,
+      "step": 10309
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5847457627118642e-05,
+      "loss": 0.0945,
+      "step": 10310
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.584510357815443e-05,
+      "loss": 0.0332,
+      "step": 10311
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5842749529190207e-05,
+      "loss": 0.1221,
+      "step": 10312
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5840395480225987e-05,
+      "loss": 0.0877,
+      "step": 10313
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.583804143126177e-05,
+      "loss": 0.0948,
+      "step": 10314
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.583568738229755e-05,
+      "loss": 0.0416,
+      "step": 10315
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5833333333333336e-05,
+      "loss": 0.12,
+      "step": 10316
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5830979284369116e-05,
+      "loss": 0.0322,
+      "step": 10317
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5828625235404897e-05,
+      "loss": 0.0577,
+      "step": 10318
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.582627118644068e-05,
+      "loss": 0.3069,
+      "step": 10319
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.582391713747646e-05,
+      "loss": 0.0328,
+      "step": 10320
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.582156308851224e-05,
+      "loss": 0.1483,
+      "step": 10321
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5819209039548026e-05,
+      "loss": 0.1365,
+      "step": 10322
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5816854990583806e-05,
+      "loss": 0.1515,
+      "step": 10323
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5814500941619583e-05,
+      "loss": 0.0493,
+      "step": 10324
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.581214689265537e-05,
+      "loss": 0.0354,
+      "step": 10325
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.580979284369115e-05,
+      "loss": 0.2904,
+      "step": 10326
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5807438794726928e-05,
+      "loss": 0.1899,
+      "step": 10327
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5805084745762716e-05,
+      "loss": 0.1062,
+      "step": 10328
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5802730696798493e-05,
+      "loss": 0.0736,
+      "step": 10329
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5800376647834273e-05,
+      "loss": 0.1842,
+      "step": 10330
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5798022598870057e-05,
+      "loss": 0.1172,
+      "step": 10331
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5795668549905838e-05,
+      "loss": 0.0768,
+      "step": 10332
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.579331450094162e-05,
+      "loss": 0.0688,
+      "step": 10333
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5790960451977402e-05,
+      "loss": 0.1026,
+      "step": 10334
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5788606403013183e-05,
+      "loss": 0.1499,
+      "step": 10335
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5786252354048967e-05,
+      "loss": 0.1154,
+      "step": 10336
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5783898305084747e-05,
+      "loss": 0.0549,
+      "step": 10337
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5781544256120528e-05,
+      "loss": 0.054,
+      "step": 10338
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.577919020715631e-05,
+      "loss": 0.0897,
+      "step": 10339
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5776836158192092e-05,
+      "loss": 0.067,
+      "step": 10340
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.577448210922787e-05,
+      "loss": 0.0542,
+      "step": 10341
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5772128060263657e-05,
+      "loss": 0.1076,
+      "step": 10342
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5769774011299437e-05,
+      "loss": 0.0421,
+      "step": 10343
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5767419962335214e-05,
+      "loss": 0.073,
+      "step": 10344
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5765065913371e-05,
+      "loss": 0.0879,
+      "step": 10345
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.576271186440678e-05,
+      "loss": 0.1161,
+      "step": 10346
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5760357815442566e-05,
+      "loss": 0.1554,
+      "step": 10347
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5758003766478343e-05,
+      "loss": 0.0426,
+      "step": 10348
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5755649717514124e-05,
+      "loss": 0.1163,
+      "step": 10349
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5753295668549908e-05,
+      "loss": 0.0948,
+      "step": 10350
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5750941619585688e-05,
+      "loss": 0.0162,
+      "step": 10351
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.574858757062147e-05,
+      "loss": 0.0453,
+      "step": 10352
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5746233521657253e-05,
+      "loss": 0.1793,
+      "step": 10353
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5743879472693033e-05,
+      "loss": 0.0277,
+      "step": 10354
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5741525423728814e-05,
+      "loss": 0.1649,
+      "step": 10355
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5739171374764598e-05,
+      "loss": 0.1743,
+      "step": 10356
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5736817325800378e-05,
+      "loss": 0.0817,
+      "step": 10357
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.573446327683616e-05,
+      "loss": 0.1127,
+      "step": 10358
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5732109227871943e-05,
+      "loss": 0.0308,
+      "step": 10359
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5729755178907723e-05,
+      "loss": 0.1161,
+      "step": 10360
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.57274011299435e-05,
+      "loss": 0.0163,
+      "step": 10361
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5725047080979288e-05,
+      "loss": 0.0788,
+      "step": 10362
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5722693032015065e-05,
+      "loss": 0.0646,
+      "step": 10363
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5720338983050852e-05,
+      "loss": 0.086,
+      "step": 10364
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.571798493408663e-05,
+      "loss": 0.1096,
+      "step": 10365
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.571563088512241e-05,
+      "loss": 0.1108,
+      "step": 10366
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5713276836158197e-05,
+      "loss": 0.1176,
+      "step": 10367
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5710922787193974e-05,
+      "loss": 0.1476,
+      "step": 10368
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5708568738229755e-05,
+      "loss": 0.1316,
+      "step": 10369
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.570621468926554e-05,
+      "loss": 0.086,
+      "step": 10370
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.570386064030132e-05,
+      "loss": 0.1388,
+      "step": 10371
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.57015065913371e-05,
+      "loss": 0.2084,
+      "step": 10372
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5699152542372884e-05,
+      "loss": 0.2041,
+      "step": 10373
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5696798493408664e-05,
+      "loss": 0.053,
+      "step": 10374
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5694444444444445e-05,
+      "loss": 0.0184,
+      "step": 10375
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.569209039548023e-05,
+      "loss": 0.1075,
+      "step": 10376
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.568973634651601e-05,
+      "loss": 0.2028,
+      "step": 10377
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5687382297551793e-05,
+      "loss": 0.1199,
+      "step": 10378
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5685028248587574e-05,
+      "loss": 0.1789,
+      "step": 10379
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.568267419962335e-05,
+      "loss": 0.0701,
+      "step": 10380
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5680320150659138e-05,
+      "loss": 0.0192,
+      "step": 10381
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5677966101694915e-05,
+      "loss": 0.0878,
+      "step": 10382
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5675612052730696e-05,
+      "loss": 0.0844,
+      "step": 10383
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5673258003766483e-05,
+      "loss": 0.1385,
+      "step": 10384
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.567090395480226e-05,
+      "loss": 0.0288,
+      "step": 10385
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.566854990583804e-05,
+      "loss": 0.151,
+      "step": 10386
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5666195856873825e-05,
+      "loss": 0.1142,
+      "step": 10387
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5663841807909605e-05,
+      "loss": 0.0737,
+      "step": 10388
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5661487758945386e-05,
+      "loss": 0.2398,
+      "step": 10389
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.565913370998117e-05,
+      "loss": 0.0648,
+      "step": 10390
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.565677966101695e-05,
+      "loss": 0.0132,
+      "step": 10391
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.565442561205273e-05,
+      "loss": 0.0602,
+      "step": 10392
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5652071563088515e-05,
+      "loss": 0.0247,
+      "step": 10393
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5649717514124295e-05,
+      "loss": 0.1266,
+      "step": 10394
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.564736346516008e-05,
+      "loss": 0.0866,
+      "step": 10395
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.564500941619586e-05,
+      "loss": 0.0613,
+      "step": 10396
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5642655367231637e-05,
+      "loss": 0.2674,
+      "step": 10397
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5640301318267424e-05,
+      "loss": 0.4012,
+      "step": 10398
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.56379472693032e-05,
+      "loss": 0.0987,
+      "step": 10399
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5635593220338982e-05,
+      "loss": 0.0544,
+      "step": 10400
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.563323917137477e-05,
+      "loss": 0.1475,
+      "step": 10401
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5630885122410546e-05,
+      "loss": 0.067,
+      "step": 10402
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5628531073446327e-05,
+      "loss": 0.1001,
+      "step": 10403
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.562617702448211e-05,
+      "loss": 0.2352,
+      "step": 10404
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.562382297551789e-05,
+      "loss": 0.0833,
+      "step": 10405
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5621468926553672e-05,
+      "loss": 0.0644,
+      "step": 10406
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5619114877589456e-05,
+      "loss": 0.1426,
+      "step": 10407
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5616760828625236e-05,
+      "loss": 0.0865,
+      "step": 10408
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5614406779661017e-05,
+      "loss": 0.0749,
+      "step": 10409
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.56120527306968e-05,
+      "loss": 0.0148,
+      "step": 10410
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.560969868173258e-05,
+      "loss": 0.09,
+      "step": 10411
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5607344632768365e-05,
+      "loss": 0.0283,
+      "step": 10412
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5604990583804146e-05,
+      "loss": 0.0461,
+      "step": 10413
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5602636534839923e-05,
+      "loss": 0.0283,
+      "step": 10414
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.560028248587571e-05,
+      "loss": 0.123,
+      "step": 10415
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5597928436911487e-05,
+      "loss": 0.1524,
+      "step": 10416
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5595574387947268e-05,
+      "loss": 0.1084,
+      "step": 10417
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5593220338983055e-05,
+      "loss": 0.1087,
+      "step": 10418
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5590866290018832e-05,
+      "loss": 0.0511,
+      "step": 10419
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5588512241054613e-05,
+      "loss": 0.0381,
+      "step": 10420
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5586158192090397e-05,
+      "loss": 0.1315,
+      "step": 10421
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5583804143126177e-05,
+      "loss": 0.0488,
+      "step": 10422
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5581450094161958e-05,
+      "loss": 0.016,
+      "step": 10423
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5579096045197742e-05,
+      "loss": 0.0438,
+      "step": 10424
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5576741996233522e-05,
+      "loss": 0.1104,
+      "step": 10425
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5574387947269306e-05,
+      "loss": 0.1661,
+      "step": 10426
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5572033898305087e-05,
+      "loss": 0.0152,
+      "step": 10427
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5569679849340867e-05,
+      "loss": 0.0926,
+      "step": 10428
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.556732580037665e-05,
+      "loss": 0.0752,
+      "step": 10429
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5564971751412432e-05,
+      "loss": 0.1885,
+      "step": 10430
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.556261770244821e-05,
+      "loss": 0.1181,
+      "step": 10431
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5560263653483996e-05,
+      "loss": 0.2665,
+      "step": 10432
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5557909604519777e-05,
+      "loss": 0.0917,
+      "step": 10433
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5555555555555554e-05,
+      "loss": 0.0775,
+      "step": 10434
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.555320150659134e-05,
+      "loss": 0.0629,
+      "step": 10435
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.555084745762712e-05,
+      "loss": 0.2529,
+      "step": 10436
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.55484934086629e-05,
+      "loss": 0.2588,
+      "step": 10437
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5546139359698683e-05,
+      "loss": 0.0556,
+      "step": 10438
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5543785310734464e-05,
+      "loss": 0.0241,
+      "step": 10439
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5541431261770244e-05,
+      "loss": 0.2712,
+      "step": 10440
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5539077212806028e-05,
+      "loss": 0.1266,
+      "step": 10441
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.553672316384181e-05,
+      "loss": 0.0699,
+      "step": 10442
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5534369114877592e-05,
+      "loss": 0.0462,
+      "step": 10443
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5532015065913373e-05,
+      "loss": 0.2407,
+      "step": 10444
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5529661016949154e-05,
+      "loss": 0.1039,
+      "step": 10445
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5527306967984937e-05,
+      "loss": 0.0979,
+      "step": 10446
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5524952919020718e-05,
+      "loss": 0.0614,
+      "step": 10447
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5522598870056495e-05,
+      "loss": 0.0794,
+      "step": 10448
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5520244821092282e-05,
+      "loss": 0.0533,
+      "step": 10449
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5517890772128063e-05,
+      "loss": 0.2952,
+      "step": 10450
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.551553672316384e-05,
+      "loss": 0.1032,
+      "step": 10451
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5513182674199627e-05,
+      "loss": 0.0333,
+      "step": 10452
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5510828625235405e-05,
+      "loss": 0.1774,
+      "step": 10453
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5508474576271185e-05,
+      "loss": 0.0507,
+      "step": 10454
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.550612052730697e-05,
+      "loss": 0.0391,
+      "step": 10455
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.550376647834275e-05,
+      "loss": 0.0958,
+      "step": 10456
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5501412429378534e-05,
+      "loss": 0.1173,
+      "step": 10457
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5499058380414314e-05,
+      "loss": 0.0515,
+      "step": 10458
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5496704331450095e-05,
+      "loss": 0.0117,
+      "step": 10459
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.549435028248588e-05,
+      "loss": 0.069,
+      "step": 10460
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.549199623352166e-05,
+      "loss": 0.1886,
+      "step": 10461
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.548964218455744e-05,
+      "loss": 0.0424,
+      "step": 10462
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5487288135593224e-05,
+      "loss": 0.0174,
+      "step": 10463
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5484934086629004e-05,
+      "loss": 0.1491,
+      "step": 10464
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.548258003766478e-05,
+      "loss": 0.1452,
+      "step": 10465
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.548022598870057e-05,
+      "loss": 0.1678,
+      "step": 10466
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.547787193973635e-05,
+      "loss": 0.0248,
+      "step": 10467
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5475517890772126e-05,
+      "loss": 0.0718,
+      "step": 10468
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5473163841807913e-05,
+      "loss": 0.0294,
+      "step": 10469
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.547080979284369e-05,
+      "loss": 0.0693,
+      "step": 10470
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.546845574387947e-05,
+      "loss": 0.0548,
+      "step": 10471
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5466101694915255e-05,
+      "loss": 0.0296,
+      "step": 10472
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5463747645951036e-05,
+      "loss": 0.0979,
+      "step": 10473
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.546139359698682e-05,
+      "loss": 0.129,
+      "step": 10474
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.54590395480226e-05,
+      "loss": 0.0559,
+      "step": 10475
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.545668549905838e-05,
+      "loss": 0.0439,
+      "step": 10476
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5454331450094165e-05,
+      "loss": 0.1013,
+      "step": 10477
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5451977401129945e-05,
+      "loss": 0.1294,
+      "step": 10478
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5449623352165726e-05,
+      "loss": 0.1739,
+      "step": 10479
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.544726930320151e-05,
+      "loss": 0.0617,
+      "step": 10480
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.544491525423729e-05,
+      "loss": 0.0191,
+      "step": 10481
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5442561205273067e-05,
+      "loss": 0.0862,
+      "step": 10482
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5440207156308855e-05,
+      "loss": 0.0508,
+      "step": 10483
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5437853107344635e-05,
+      "loss": 0.1686,
+      "step": 10484
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5435499058380412e-05,
+      "loss": 0.0812,
+      "step": 10485
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.54331450094162e-05,
+      "loss": 0.0585,
+      "step": 10486
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5430790960451977e-05,
+      "loss": 0.1106,
+      "step": 10487
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5428436911487757e-05,
+      "loss": 0.0673,
+      "step": 10488
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.542608286252354e-05,
+      "loss": 0.0354,
+      "step": 10489
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5423728813559322e-05,
+      "loss": 0.0473,
+      "step": 10490
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5421374764595106e-05,
+      "loss": 0.027,
+      "step": 10491
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5419020715630886e-05,
+      "loss": 0.0756,
+      "step": 10492
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5416666666666667e-05,
+      "loss": 0.053,
+      "step": 10493
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.541431261770245e-05,
+      "loss": 0.2423,
+      "step": 10494
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.541195856873823e-05,
+      "loss": 0.1164,
+      "step": 10495
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5409604519774012e-05,
+      "loss": 0.1179,
+      "step": 10496
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5407250470809796e-05,
+      "loss": 0.0456,
+      "step": 10497
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5404896421845576e-05,
+      "loss": 0.0162,
+      "step": 10498
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5402542372881353e-05,
+      "loss": 0.0786,
+      "step": 10499
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.540018832391714e-05,
+      "loss": 0.0314,
+      "step": 10500
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.539783427495292e-05,
+      "loss": 0.1295,
+      "step": 10501
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5395480225988698e-05,
+      "loss": 0.269,
+      "step": 10502
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5393126177024486e-05,
+      "loss": 0.1797,
+      "step": 10503
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5390772128060263e-05,
+      "loss": 0.2424,
+      "step": 10504
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.538841807909605e-05,
+      "loss": 0.0874,
+      "step": 10505
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5386064030131827e-05,
+      "loss": 0.0212,
+      "step": 10506
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5383709981167608e-05,
+      "loss": 0.1963,
+      "step": 10507
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.538135593220339e-05,
+      "loss": 0.1436,
+      "step": 10508
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5379001883239172e-05,
+      "loss": 0.1728,
+      "step": 10509
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5376647834274953e-05,
+      "loss": 0.1044,
+      "step": 10510
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5374293785310737e-05,
+      "loss": 0.1186,
+      "step": 10511
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5371939736346517e-05,
+      "loss": 0.0287,
+      "step": 10512
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5369585687382298e-05,
+      "loss": 0.11,
+      "step": 10513
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.536723163841808e-05,
+      "loss": 0.0759,
+      "step": 10514
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5364877589453862e-05,
+      "loss": 0.2138,
+      "step": 10515
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5362523540489643e-05,
+      "loss": 0.0558,
+      "step": 10516
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5360169491525427e-05,
+      "loss": 0.0988,
+      "step": 10517
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5357815442561207e-05,
+      "loss": 0.1043,
+      "step": 10518
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5355461393596984e-05,
+      "loss": 0.1266,
+      "step": 10519
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.535310734463277e-05,
+      "loss": 0.1827,
+      "step": 10520
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.535075329566855e-05,
+      "loss": 0.123,
+      "step": 10521
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5348399246704336e-05,
+      "loss": 0.0643,
+      "step": 10522
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5346045197740113e-05,
+      "loss": 0.0159,
+      "step": 10523
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5343691148775894e-05,
+      "loss": 0.0946,
+      "step": 10524
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.534133709981168e-05,
+      "loss": 0.1894,
+      "step": 10525
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5338983050847458e-05,
+      "loss": 0.0931,
+      "step": 10526
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.533662900188324e-05,
+      "loss": 0.1888,
+      "step": 10527
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5334274952919023e-05,
+      "loss": 0.1702,
+      "step": 10528
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5331920903954803e-05,
+      "loss": 0.0858,
+      "step": 10529
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5329566854990584e-05,
+      "loss": 0.0375,
+      "step": 10530
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5327212806026368e-05,
+      "loss": 0.2063,
+      "step": 10531
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5324858757062148e-05,
+      "loss": 0.0373,
+      "step": 10532
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.532250470809793e-05,
+      "loss": 0.1197,
+      "step": 10533
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5320150659133713e-05,
+      "loss": 0.0767,
+      "step": 10534
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5317796610169493e-05,
+      "loss": 0.0606,
+      "step": 10535
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5315442561205277e-05,
+      "loss": 0.0203,
+      "step": 10536
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5313088512241058e-05,
+      "loss": 0.0647,
+      "step": 10537
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5310734463276835e-05,
+      "loss": 0.0967,
+      "step": 10538
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5308380414312622e-05,
+      "loss": 0.1549,
+      "step": 10539
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.53060263653484e-05,
+      "loss": 0.0628,
+      "step": 10540
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.530367231638418e-05,
+      "loss": 0.0938,
+      "step": 10541
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5301318267419967e-05,
+      "loss": 0.216,
+      "step": 10542
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5298964218455744e-05,
+      "loss": 0.0965,
+      "step": 10543
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5296610169491525e-05,
+      "loss": 0.1336,
+      "step": 10544
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.529425612052731e-05,
+      "loss": 0.1048,
+      "step": 10545
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.529190207156309e-05,
+      "loss": 0.0841,
+      "step": 10546
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.528954802259887e-05,
+      "loss": 0.0896,
+      "step": 10547
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5287193973634654e-05,
+      "loss": 0.1985,
+      "step": 10548
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5284839924670434e-05,
+      "loss": 0.059,
+      "step": 10549
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5282485875706215e-05,
+      "loss": 0.0728,
+      "step": 10550
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5280131826742e-05,
+      "loss": 0.114,
+      "step": 10551
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.527777777777778e-05,
+      "loss": 0.0295,
+      "step": 10552
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5275423728813563e-05,
+      "loss": 0.0849,
+      "step": 10553
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5273069679849344e-05,
+      "loss": 0.1484,
+      "step": 10554
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.527071563088512e-05,
+      "loss": 0.1272,
+      "step": 10555
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5268361581920908e-05,
+      "loss": 0.3028,
+      "step": 10556
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5266007532956685e-05,
+      "loss": 0.0862,
+      "step": 10557
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5263653483992466e-05,
+      "loss": 0.0897,
+      "step": 10558
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5261299435028253e-05,
+      "loss": 0.0719,
+      "step": 10559
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.525894538606403e-05,
+      "loss": 0.0581,
+      "step": 10560
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.525659133709981e-05,
+      "loss": 0.0792,
+      "step": 10561
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5254237288135595e-05,
+      "loss": 0.0493,
+      "step": 10562
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5251883239171375e-05,
+      "loss": 0.1273,
+      "step": 10563
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5249529190207156e-05,
+      "loss": 0.1364,
+      "step": 10564
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.524717514124294e-05,
+      "loss": 0.019,
+      "step": 10565
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.524482109227872e-05,
+      "loss": 0.1679,
+      "step": 10566
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.52424670433145e-05,
+      "loss": 0.175,
+      "step": 10567
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5240112994350285e-05,
+      "loss": 0.3573,
+      "step": 10568
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5237758945386065e-05,
+      "loss": 0.0249,
+      "step": 10569
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.523540489642185e-05,
+      "loss": 0.0946,
+      "step": 10570
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.523305084745763e-05,
+      "loss": 0.02,
+      "step": 10571
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5230696798493407e-05,
+      "loss": 0.1789,
+      "step": 10572
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5228342749529194e-05,
+      "loss": 0.1108,
+      "step": 10573
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.522598870056497e-05,
+      "loss": 0.1708,
+      "step": 10574
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5223634651600752e-05,
+      "loss": 0.0663,
+      "step": 10575
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.522128060263654e-05,
+      "loss": 0.0233,
+      "step": 10576
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5218926553672316e-05,
+      "loss": 0.0915,
+      "step": 10577
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5216572504708097e-05,
+      "loss": 0.0842,
+      "step": 10578
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.521421845574388e-05,
+      "loss": 0.117,
+      "step": 10579
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.521186440677966e-05,
+      "loss": 0.0997,
+      "step": 10580
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5209510357815442e-05,
+      "loss": 0.1619,
+      "step": 10581
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5207156308851226e-05,
+      "loss": 0.0481,
+      "step": 10582
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5204802259887006e-05,
+      "loss": 0.0732,
+      "step": 10583
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.520244821092279e-05,
+      "loss": 0.1284,
+      "step": 10584
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.520009416195857e-05,
+      "loss": 0.083,
+      "step": 10585
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.519774011299435e-05,
+      "loss": 0.0804,
+      "step": 10586
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5195386064030135e-05,
+      "loss": 0.0899,
+      "step": 10587
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5193032015065916e-05,
+      "loss": 0.097,
+      "step": 10588
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5190677966101693e-05,
+      "loss": 0.1378,
+      "step": 10589
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.518832391713748e-05,
+      "loss": 0.055,
+      "step": 10590
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.518596986817326e-05,
+      "loss": 0.1583,
+      "step": 10591
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5183615819209038e-05,
+      "loss": 0.1909,
+      "step": 10592
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5181261770244825e-05,
+      "loss": 0.2236,
+      "step": 10593
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5178907721280603e-05,
+      "loss": 0.041,
+      "step": 10594
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5176553672316383e-05,
+      "loss": 0.0584,
+      "step": 10595
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5174199623352167e-05,
+      "loss": 0.0993,
+      "step": 10596
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5171845574387948e-05,
+      "loss": 0.1445,
+      "step": 10597
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5169491525423728e-05,
+      "loss": 0.0852,
+      "step": 10598
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5167137476459512e-05,
+      "loss": 0.061,
+      "step": 10599
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5164783427495293e-05,
+      "loss": 0.0654,
+      "step": 10600
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5162429378531076e-05,
+      "loss": 0.0774,
+      "step": 10601
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5160075329566857e-05,
+      "loss": 0.0749,
+      "step": 10602
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5157721280602638e-05,
+      "loss": 0.1631,
+      "step": 10603
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.515536723163842e-05,
+      "loss": 0.1183,
+      "step": 10604
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5153013182674202e-05,
+      "loss": 0.0496,
+      "step": 10605
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.515065913370998e-05,
+      "loss": 0.0292,
+      "step": 10606
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5148305084745766e-05,
+      "loss": 0.0704,
+      "step": 10607
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5145951035781547e-05,
+      "loss": 0.0739,
+      "step": 10608
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5143596986817324e-05,
+      "loss": 0.0841,
+      "step": 10609
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.514124293785311e-05,
+      "loss": 0.0155,
+      "step": 10610
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.513888888888889e-05,
+      "loss": 0.1113,
+      "step": 10611
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.513653483992467e-05,
+      "loss": 0.1875,
+      "step": 10612
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5134180790960453e-05,
+      "loss": 0.0591,
+      "step": 10613
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5131826741996234e-05,
+      "loss": 0.0944,
+      "step": 10614
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5129472693032017e-05,
+      "loss": 0.0562,
+      "step": 10615
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5127118644067798e-05,
+      "loss": 0.0391,
+      "step": 10616
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.512476459510358e-05,
+      "loss": 0.1603,
+      "step": 10617
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5122410546139362e-05,
+      "loss": 0.058,
+      "step": 10618
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5120056497175143e-05,
+      "loss": 0.0359,
+      "step": 10619
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5117702448210924e-05,
+      "loss": 0.0889,
+      "step": 10620
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5115348399246707e-05,
+      "loss": 0.1261,
+      "step": 10621
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5112994350282488e-05,
+      "loss": 0.0425,
+      "step": 10622
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5110640301318265e-05,
+      "loss": 0.1107,
+      "step": 10623
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5108286252354052e-05,
+      "loss": 0.1409,
+      "step": 10624
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5105932203389833e-05,
+      "loss": 0.1573,
+      "step": 10625
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.510357815442561e-05,
+      "loss": 0.0213,
+      "step": 10626
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5101224105461397e-05,
+      "loss": 0.0254,
+      "step": 10627
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5098870056497175e-05,
+      "loss": 0.0471,
+      "step": 10628
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5096516007532955e-05,
+      "loss": 0.0808,
+      "step": 10629
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.509416195856874e-05,
+      "loss": 0.0452,
+      "step": 10630
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.509180790960452e-05,
+      "loss": 0.1359,
+      "step": 10631
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5089453860640304e-05,
+      "loss": 0.0757,
+      "step": 10632
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5087099811676084e-05,
+      "loss": 0.0831,
+      "step": 10633
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5084745762711865e-05,
+      "loss": 0.0934,
+      "step": 10634
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.508239171374765e-05,
+      "loss": 0.0608,
+      "step": 10635
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.508003766478343e-05,
+      "loss": 0.1007,
+      "step": 10636
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.507768361581921e-05,
+      "loss": 0.1757,
+      "step": 10637
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5075329566854994e-05,
+      "loss": 0.018,
+      "step": 10638
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5072975517890774e-05,
+      "loss": 0.0298,
+      "step": 10639
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.507062146892655e-05,
+      "loss": 0.1424,
+      "step": 10640
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.506826741996234e-05,
+      "loss": 0.1086,
+      "step": 10641
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.506591337099812e-05,
+      "loss": 0.0548,
+      "step": 10642
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5063559322033896e-05,
+      "loss": 0.113,
+      "step": 10643
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5061205273069684e-05,
+      "loss": 0.0489,
+      "step": 10644
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.505885122410546e-05,
+      "loss": 0.0586,
+      "step": 10645
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.902651748744604,
+      "eval_loss": 0.3379899263381958,
+      "eval_runtime": 22.2851,
+      "eval_samples_per_second": 1528.063,
+      "eval_steps_per_second": 23.917,
+      "step": 10645
+    }
+  ],
+  "max_steps": 21290,
+  "num_train_epochs": 10,
+  "total_flos": 6889551786482160.0,
+  "trial_name": null,
+  "trial_params": null
+}