diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,18024 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.1,
+  "global_step": 3000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 1.6356,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 1.569,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 1.4907,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 1.6294,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 1.5081,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 1.5219,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 1.7132,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 1.6568,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-07,
+      "loss": 1.4473,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 1.404,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1e-06,
+      "loss": 1.7641,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 1.5949,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3e-06,
+      "loss": 1.5707,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 1.556,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-06,
+      "loss": 1.2753,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 1.5217,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 1.4203,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-06,
+      "loss": 1.6379,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9e-06,
+      "loss": 1.4178,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.4491,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 1.2947,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2e-06,
+      "loss": 1.3568,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3e-06,
+      "loss": 1.4005,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 1.3477,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-06,
+      "loss": 1.257,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6e-06,
+      "loss": 1.2816,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7e-06,
+      "loss": 1.4465,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 1.3507,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9e-06,
+      "loss": 1.2533,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-06,
+      "loss": 1.2971,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1e-06,
+      "loss": 1.5496,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 1.3907,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3e-06,
+      "loss": 1.3477,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 1.365,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5000000000000004e-06,
+      "loss": 1.2424,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6e-06,
+      "loss": 1.3346,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7e-06,
+      "loss": 1.1961,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8e-06,
+      "loss": 1.1934,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9e-06,
+      "loss": 1.4738,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.3204,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1000000000000006e-06,
+      "loss": 1.1442,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 1.2211,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2999999999999995e-06,
+      "loss": 1.3181,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4e-06,
+      "loss": 1.2945,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5e-06,
+      "loss": 1.3147,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6e-06,
+      "loss": 1.1463,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7e-06,
+      "loss": 1.2946,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 1.2329,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 1.2244,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 1.1023,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.1e-06,
+      "loss": 1.1198,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.2e-06,
+      "loss": 1.0063,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.3e-06,
+      "loss": 1.286,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.4e-06,
+      "loss": 1.1782,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 1.1989,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 0.9778,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.7000000000000005e-06,
+      "loss": 1.124,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.8e-06,
+      "loss": 1.1595,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9e-06,
+      "loss": 1.0188,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-06,
+      "loss": 1.215,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.1e-06,
+      "loss": 0.9485,
+      "step": 61
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.2e-06,
+      "loss": 1.0355,
+      "step": 62
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.300000000000001e-06,
+      "loss": 1.2669,
+      "step": 63
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 1.1419,
+      "step": 64
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 0.982,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.6e-06,
+      "loss": 1.0798,
+      "step": 66
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.700000000000001e-06,
+      "loss": 0.9395,
+      "step": 67
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 1.0678,
+      "step": 68
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.900000000000001e-06,
+      "loss": 1.0312,
+      "step": 69
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 1.0835,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.1e-06,
+      "loss": 1.0525,
+      "step": 71
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.2e-06,
+      "loss": 1.0649,
+      "step": 72
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.2999999999999996e-06,
+      "loss": 0.9691,
+      "step": 73
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.4e-06,
+      "loss": 0.9828,
+      "step": 74
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.5e-06,
+      "loss": 1.0309,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.6e-06,
+      "loss": 0.9565,
+      "step": 76
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.7e-06,
+      "loss": 1.0238,
+      "step": 77
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.8e-06,
+      "loss": 1.1796,
+      "step": 78
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.9e-06,
+      "loss": 1.0269,
+      "step": 79
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.92,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.1e-06,
+      "loss": 1.0237,
+      "step": 81
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.200000000000001e-06,
+      "loss": 0.8151,
+      "step": 82
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.3e-06,
+      "loss": 0.9815,
+      "step": 83
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 0.9809,
+      "step": 84
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.500000000000002e-06,
+      "loss": 0.9286,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.599999999999999e-06,
+      "loss": 0.9134,
+      "step": 86
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.7e-06,
+      "loss": 1.0288,
+      "step": 87
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.8e-06,
+      "loss": 0.9538,
+      "step": 88
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.9e-06,
+      "loss": 0.7878,
+      "step": 89
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-06,
+      "loss": 1.0201,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.100000000000001e-06,
+      "loss": 0.9398,
+      "step": 91
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.2e-06,
+      "loss": 0.8696,
+      "step": 92
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.3e-06,
+      "loss": 1.0257,
+      "step": 93
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.4e-06,
+      "loss": 0.9946,
+      "step": 94
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.5e-06,
+      "loss": 0.9029,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 1.1029,
+      "step": 96
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.7e-06,
+      "loss": 0.9285,
+      "step": 97
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 0.97,
+      "step": 98
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.900000000000002e-06,
+      "loss": 0.7951,
+      "step": 99
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 1.0978,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0100000000000002e-05,
+      "loss": 0.9044,
+      "step": 101
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.02e-05,
+      "loss": 0.8795,
+      "step": 102
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.03e-05,
+      "loss": 0.9367,
+      "step": 103
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.04e-05,
+      "loss": 0.8666,
+      "step": 104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.05e-05,
+      "loss": 0.9539,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.06e-05,
+      "loss": 0.9758,
+      "step": 106
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0700000000000001e-05,
+      "loss": 1.012,
+      "step": 107
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.08e-05,
+      "loss": 0.9424,
+      "step": 108
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.09e-05,
+      "loss": 0.9445,
+      "step": 109
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 1.0557,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.11e-05,
+      "loss": 0.9198,
+      "step": 111
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 1.0217,
+      "step": 112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.13e-05,
+      "loss": 0.7384,
+      "step": 113
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 0.7724,
+      "step": 114
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1500000000000002e-05,
+      "loss": 0.9663,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.16e-05,
+      "loss": 1.001,
+      "step": 116
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1700000000000001e-05,
+      "loss": 0.8286,
+      "step": 117
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.18e-05,
+      "loss": 0.9012,
+      "step": 118
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.19e-05,
+      "loss": 0.9115,
+      "step": 119
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-05,
+      "loss": 0.8661,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2100000000000001e-05,
+      "loss": 0.8695,
+      "step": 121
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.22e-05,
+      "loss": 1.0116,
+      "step": 122
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.23e-05,
+      "loss": 0.8426,
+      "step": 123
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.24e-05,
+      "loss": 0.9811,
+      "step": 124
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.25e-05,
+      "loss": 0.8201,
+      "step": 125
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2600000000000001e-05,
+      "loss": 0.7438,
+      "step": 126
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.27e-05,
+      "loss": 0.8961,
+      "step": 127
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 0.8884,
+      "step": 128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.29e-05,
+      "loss": 0.9154,
+      "step": 129
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 0.8441,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3100000000000002e-05,
+      "loss": 0.9825,
+      "step": 131
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.32e-05,
+      "loss": 0.9651,
+      "step": 132
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3300000000000001e-05,
+      "loss": 0.8907,
+      "step": 133
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3400000000000002e-05,
+      "loss": 1.1089,
+      "step": 134
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 1.1788,
+      "step": 135
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 0.9596,
+      "step": 136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3700000000000001e-05,
+      "loss": 0.9166,
+      "step": 137
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 0.849,
+      "step": 138
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3900000000000002e-05,
+      "loss": 0.8452,
+      "step": 139
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 0.9187,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4099999999999999e-05,
+      "loss": 0.9624,
+      "step": 141
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.42e-05,
+      "loss": 0.8663,
+      "step": 142
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.43e-05,
+      "loss": 0.8919,
+      "step": 143
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.44e-05,
+      "loss": 0.8292,
+      "step": 144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.45e-05,
+      "loss": 0.9521,
+      "step": 145
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4599999999999999e-05,
+      "loss": 0.9328,
+      "step": 146
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.47e-05,
+      "loss": 0.9945,
+      "step": 147
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.48e-05,
+      "loss": 0.9507,
+      "step": 148
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.49e-05,
+      "loss": 1.0561,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5e-05,
+      "loss": 0.8099,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.51e-05,
+      "loss": 0.8593,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.52e-05,
+      "loss": 0.9603,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.53e-05,
+      "loss": 0.8324,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.54e-05,
+      "loss": 0.8616,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.55e-05,
+      "loss": 0.8296,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.56e-05,
+      "loss": 0.9316,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5700000000000002e-05,
+      "loss": 1.0482,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.58e-05,
+      "loss": 0.833,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.59e-05,
+      "loss": 0.971,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.7481,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6100000000000002e-05,
+      "loss": 0.905,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.62e-05,
+      "loss": 0.7904,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.63e-05,
+      "loss": 0.9402,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6400000000000002e-05,
+      "loss": 0.9418,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.65e-05,
+      "loss": 1.0416,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.66e-05,
+      "loss": 1.0899,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6700000000000003e-05,
+      "loss": 0.8025,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 0.7448,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.69e-05,
+      "loss": 0.9167,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 0.8892,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7100000000000002e-05,
+      "loss": 0.8696,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7199999999999998e-05,
+      "loss": 0.6993,
+      "step": 172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.73e-05,
+      "loss": 0.81,
+      "step": 173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.74e-05,
+      "loss": 0.9277,
+      "step": 174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.75e-05,
+      "loss": 0.9119,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.76e-05,
+      "loss": 0.8864,
+      "step": 176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.77e-05,
+      "loss": 0.9153,
+      "step": 177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.78e-05,
+      "loss": 0.7472,
+      "step": 178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.79e-05,
+      "loss": 0.7942,
+      "step": 179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8e-05,
+      "loss": 0.8487,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.81e-05,
+      "loss": 0.8524,
+      "step": 181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8200000000000002e-05,
+      "loss": 1.0672,
+      "step": 182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.83e-05,
+      "loss": 1.1653,
+      "step": 183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.84e-05,
+      "loss": 0.7624,
+      "step": 184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.85e-05,
+      "loss": 0.9921,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.86e-05,
+      "loss": 0.791,
+      "step": 186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.87e-05,
+      "loss": 0.9381,
+      "step": 187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.88e-05,
+      "loss": 0.8747,
+      "step": 188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8900000000000002e-05,
+      "loss": 0.8154,
+      "step": 189
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9e-05,
+      "loss": 0.7995,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.91e-05,
+      "loss": 0.9937,
+      "step": 191
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 0.8043,
+      "step": 192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.93e-05,
+      "loss": 0.7926,
+      "step": 193
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.94e-05,
+      "loss": 0.8971,
+      "step": 194
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9500000000000003e-05,
+      "loss": 0.8426,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 0.7577,
+      "step": 196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.97e-05,
+      "loss": 0.9891,
+      "step": 197
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9800000000000004e-05,
+      "loss": 0.8518,
+      "step": 198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9900000000000003e-05,
+      "loss": 0.8523,
+      "step": 199
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.9556,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.01e-05,
+      "loss": 0.8584,
+      "step": 201
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0200000000000003e-05,
+      "loss": 0.72,
+      "step": 202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0300000000000002e-05,
+      "loss": 0.9775,
+      "step": 203
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.04e-05,
+      "loss": 0.8237,
+      "step": 204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.05e-05,
+      "loss": 0.8479,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.06e-05,
+      "loss": 0.7963,
+      "step": 206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.07e-05,
+      "loss": 0.7799,
+      "step": 207
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.08e-05,
+      "loss": 0.7951,
+      "step": 208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.09e-05,
+      "loss": 0.971,
+      "step": 209
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1e-05,
+      "loss": 0.859,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.11e-05,
+      "loss": 0.812,
+      "step": 211
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.12e-05,
+      "loss": 0.7966,
+      "step": 212
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.13e-05,
+      "loss": 0.8498,
+      "step": 213
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1400000000000002e-05,
+      "loss": 0.6947,
+      "step": 214
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.15e-05,
+      "loss": 0.8183,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.16e-05,
+      "loss": 0.7281,
+      "step": 216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1700000000000002e-05,
+      "loss": 0.8191,
+      "step": 217
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.18e-05,
+      "loss": 0.9327,
+      "step": 218
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.19e-05,
+      "loss": 0.6455,
+      "step": 219
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 0.7371,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2100000000000002e-05,
+      "loss": 0.6995,
+      "step": 221
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.22e-05,
+      "loss": 1.0339,
+      "step": 222
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.23e-05,
+      "loss": 0.7815,
+      "step": 223
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2400000000000002e-05,
+      "loss": 0.838,
+      "step": 224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.25e-05,
+      "loss": 0.8821,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.26e-05,
+      "loss": 0.8844,
+      "step": 226
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2700000000000003e-05,
+      "loss": 0.7598,
+      "step": 227
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 0.9085,
+      "step": 228
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.29e-05,
+      "loss": 0.7889,
+      "step": 229
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 0.8648,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3100000000000002e-05,
+      "loss": 0.9301,
+      "step": 231
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.32e-05,
+      "loss": 0.6923,
+      "step": 232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3300000000000004e-05,
+      "loss": 0.8027,
+      "step": 233
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3400000000000003e-05,
+      "loss": 0.789,
+      "step": 234
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.35e-05,
+      "loss": 0.7242,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.36e-05,
+      "loss": 0.7613,
+      "step": 236
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.37e-05,
+      "loss": 0.7279,
+      "step": 237
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.38e-05,
+      "loss": 0.7436,
+      "step": 238
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.39e-05,
+      "loss": 0.7942,
+      "step": 239
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4e-05,
+      "loss": 0.7837,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.41e-05,
+      "loss": 0.7857,
+      "step": 241
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4200000000000002e-05,
+      "loss": 0.7404,
+      "step": 242
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.43e-05,
+      "loss": 0.7993,
+      "step": 243
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.44e-05,
+      "loss": 0.6911,
+      "step": 244
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.45e-05,
+      "loss": 0.7615,
+      "step": 245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.46e-05,
+      "loss": 0.8835,
+      "step": 246
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.47e-05,
+      "loss": 0.7773,
+      "step": 247
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.48e-05,
+      "loss": 0.6872,
+      "step": 248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4900000000000002e-05,
+      "loss": 0.7611,
+      "step": 249
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-05,
+      "loss": 0.6354,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.51e-05,
+      "loss": 0.7574,
+      "step": 251
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5200000000000003e-05,
+      "loss": 0.7275,
+      "step": 252
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5300000000000002e-05,
+      "loss": 0.7098,
+      "step": 253
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.54e-05,
+      "loss": 0.7069,
+      "step": 254
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5500000000000003e-05,
+      "loss": 0.8635,
+      "step": 255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5600000000000002e-05,
+      "loss": 0.7811,
+      "step": 256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.57e-05,
+      "loss": 0.9526,
+      "step": 257
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.58e-05,
+      "loss": 0.7802,
+      "step": 258
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5900000000000003e-05,
+      "loss": 0.7543,
+      "step": 259
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 0.8365,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.61e-05,
+      "loss": 0.715,
+      "step": 261
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6200000000000003e-05,
+      "loss": 0.6392,
+      "step": 262
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6300000000000002e-05,
+      "loss": 0.7498,
+      "step": 263
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.64e-05,
+      "loss": 0.7108,
+      "step": 264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6500000000000004e-05,
+      "loss": 0.7455,
+      "step": 265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6600000000000003e-05,
+      "loss": 0.9173,
+      "step": 266
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6700000000000002e-05,
+      "loss": 0.8534,
+      "step": 267
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6800000000000004e-05,
+      "loss": 0.6896,
+      "step": 268
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6900000000000003e-05,
+      "loss": 0.6906,
+      "step": 269
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 0.8281,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7100000000000005e-05,
+      "loss": 0.7837,
+      "step": 271
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7200000000000004e-05,
+      "loss": 0.672,
+      "step": 272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7300000000000003e-05,
+      "loss": 0.7818,
+      "step": 273
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7400000000000002e-05,
+      "loss": 0.8296,
+      "step": 274
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7500000000000004e-05,
+      "loss": 0.7745,
+      "step": 275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7600000000000003e-05,
+      "loss": 1.0863,
+      "step": 276
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7700000000000002e-05,
+      "loss": 0.8462,
+      "step": 277
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7800000000000005e-05,
+      "loss": 0.8386,
+      "step": 278
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7900000000000004e-05,
+      "loss": 0.6764,
+      "step": 279
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.6635,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8100000000000005e-05,
+      "loss": 0.8693,
+      "step": 281
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 0.8501,
+      "step": 282
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.83e-05,
+      "loss": 0.5798,
+      "step": 283
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.84e-05,
+      "loss": 0.8188,
+      "step": 284
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 1.1554,
+      "step": 285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.86e-05,
+      "loss": 0.7327,
+      "step": 286
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.87e-05,
+      "loss": 0.8776,
+      "step": 287
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.88e-05,
+      "loss": 0.7753,
+      "step": 288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8899999999999998e-05,
+      "loss": 0.5591,
+      "step": 289
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9e-05,
+      "loss": 0.6504,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.91e-05,
+      "loss": 0.6921,
+      "step": 291
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9199999999999998e-05,
+      "loss": 0.9494,
+      "step": 292
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.93e-05,
+      "loss": 0.6978,
+      "step": 293
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.94e-05,
+      "loss": 0.8789,
+      "step": 294
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.95e-05,
+      "loss": 0.9384,
+      "step": 295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.96e-05,
+      "loss": 0.749,
+      "step": 296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.97e-05,
+      "loss": 0.7065,
+      "step": 297
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.98e-05,
+      "loss": 0.8325,
+      "step": 298
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9900000000000002e-05,
+      "loss": 0.7789,
+      "step": 299
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-05,
+      "loss": 0.6741,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.01e-05,
+      "loss": 0.6289,
+      "step": 301
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.02e-05,
+      "loss": 0.651,
+      "step": 302
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.03e-05,
+      "loss": 0.7833,
+      "step": 303
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.04e-05,
+      "loss": 0.8017,
+      "step": 304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.05e-05,
+      "loss": 0.5588,
+      "step": 305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.06e-05,
+      "loss": 1.1015,
+      "step": 306
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.07e-05,
+      "loss": 0.8201,
+      "step": 307
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.08e-05,
+      "loss": 0.6623,
+      "step": 308
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.09e-05,
+      "loss": 0.6466,
+      "step": 309
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1e-05,
+      "loss": 0.815,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1100000000000004e-05,
+      "loss": 0.6209,
+      "step": 311
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.12e-05,
+      "loss": 0.6524,
+      "step": 312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.13e-05,
+      "loss": 0.674,
+      "step": 313
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1400000000000004e-05,
+      "loss": 0.7062,
+      "step": 314
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.15e-05,
+      "loss": 0.7245,
+      "step": 315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.16e-05,
+      "loss": 0.6551,
+      "step": 316
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1700000000000005e-05,
+      "loss": 0.6009,
+      "step": 317
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.18e-05,
+      "loss": 0.7824,
+      "step": 318
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.19e-05,
+      "loss": 0.7887,
+      "step": 319
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 1.3951,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.21e-05,
+      "loss": 2.443,
+      "step": 321
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2200000000000003e-05,
+      "loss": 1.8619,
+      "step": 322
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2300000000000006e-05,
+      "loss": 2.1553,
+      "step": 323
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.24e-05,
+      "loss": 1.2818,
+      "step": 324
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2500000000000004e-05,
+      "loss": 2.1438,
+      "step": 325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.26e-05,
+      "loss": 1.692,
+      "step": 326
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.27e-05,
+      "loss": 1.9262,
+      "step": 327
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2800000000000004e-05,
+      "loss": 1.1218,
+      "step": 328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.29e-05,
+      "loss": 1.0652,
+      "step": 329
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3e-05,
+      "loss": 2.0552,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3100000000000005e-05,
+      "loss": 2.2341,
+      "step": 331
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.32e-05,
+      "loss": 0.966,
+      "step": 332
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.33e-05,
+      "loss": 0.7042,
+      "step": 333
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3400000000000005e-05,
+      "loss": 1.3103,
+      "step": 334
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.35e-05,
+      "loss": 2.6564,
+      "step": 335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3600000000000004e-05,
+      "loss": 1.1635,
+      "step": 336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3700000000000006e-05,
+      "loss": 2.955,
+      "step": 337
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.38e-05,
+      "loss": 1.1058,
+      "step": 338
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3900000000000004e-05,
+      "loss": 1.6789,
+      "step": 339
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 1.8541,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.41e-05,
+      "loss": 2.5678,
+      "step": 341
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4200000000000005e-05,
+      "loss": 1.6021,
+      "step": 342
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.430000000000001e-05,
+      "loss": 1.6675,
+      "step": 343
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4399999999999996e-05,
+      "loss": 2.1323,
+      "step": 344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.45e-05,
+      "loss": 1.6304,
+      "step": 345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.46e-05,
+      "loss": 1.4361,
+      "step": 346
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4699999999999996e-05,
+      "loss": 1.6252,
+      "step": 347
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.48e-05,
+      "loss": 2.0106,
+      "step": 348
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.49e-05,
+      "loss": 1.3912,
+      "step": 349
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5e-05,
+      "loss": 1.0489,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.51e-05,
+      "loss": 1.548,
+      "step": 351
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.52e-05,
+      "loss": 1.0265,
+      "step": 352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.53e-05,
+      "loss": 2.6307,
+      "step": 353
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.54e-05,
+      "loss": 2.303,
+      "step": 354
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.55e-05,
+      "loss": 2.2019,
+      "step": 355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.56e-05,
+      "loss": 1.3886,
+      "step": 356
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.57e-05,
+      "loss": 1.3356,
+      "step": 357
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.58e-05,
+      "loss": 1.3692,
+      "step": 358
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.59e-05,
+      "loss": 1.7251,
+      "step": 359
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6e-05,
+      "loss": 1.7248,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.61e-05,
+      "loss": 1.9716,
+      "step": 361
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.62e-05,
+      "loss": 1.2527,
+      "step": 362
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.63e-05,
+      "loss": 1.98,
+      "step": 363
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6400000000000004e-05,
+      "loss": 1.5317,
+      "step": 364
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.65e-05,
+      "loss": 2.0971,
+      "step": 365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.66e-05,
+      "loss": 1.2755,
+      "step": 366
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6700000000000004e-05,
+      "loss": 1.6604,
+      "step": 367
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.68e-05,
+      "loss": 0.9964,
+      "step": 368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.69e-05,
+      "loss": 1.6341,
+      "step": 369
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7e-05,
+      "loss": 2.0033,
+      "step": 370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.71e-05,
+      "loss": 1.3873,
+      "step": 371
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.72e-05,
+      "loss": 2.1517,
+      "step": 372
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.73e-05,
+      "loss": 1.3119,
+      "step": 373
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.74e-05,
+      "loss": 1.9879,
+      "step": 374
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 1.3931,
+      "step": 375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.76e-05,
+      "loss": 1.0887,
+      "step": 376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.77e-05,
+      "loss": 1.3745,
+      "step": 377
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7800000000000004e-05,
+      "loss": 1.3452,
+      "step": 378
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.79e-05,
+      "loss": 1.6338,
+      "step": 379
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8e-05,
+      "loss": 1.0496,
+      "step": 380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8100000000000005e-05,
+      "loss": 1.2968,
+      "step": 381
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.82e-05,
+      "loss": 2.8746,
+      "step": 382
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.83e-05,
+      "loss": 1.3141,
+      "step": 383
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8400000000000005e-05,
+      "loss": 1.9839,
+      "step": 384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.85e-05,
+      "loss": 1.9305,
+      "step": 385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.86e-05,
+      "loss": 1.8477,
+      "step": 386
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8700000000000006e-05,
+      "loss": 2.1424,
+      "step": 387
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.88e-05,
+      "loss": 1.9054,
+      "step": 388
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8900000000000004e-05,
+      "loss": 1.6616,
+      "step": 389
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 0.8488,
+      "step": 390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.91e-05,
+      "loss": 1.9558,
+      "step": 391
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9200000000000004e-05,
+      "loss": 1.4621,
+      "step": 392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9300000000000007e-05,
+      "loss": 0.6847,
+      "step": 393
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.94e-05,
+      "loss": 0.6939,
+      "step": 394
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9500000000000005e-05,
+      "loss": 0.7112,
+      "step": 395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.960000000000001e-05,
+      "loss": 0.6505,
+      "step": 396
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.97e-05,
+      "loss": 0.8862,
+      "step": 397
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9800000000000005e-05,
+      "loss": 0.8216,
+      "step": 398
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.99e-05,
+      "loss": 0.8588,
+      "step": 399
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4e-05,
+      "loss": 0.6358,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0100000000000006e-05,
+      "loss": 0.6348,
+      "step": 401
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.02e-05,
+      "loss": 0.9504,
+      "step": 402
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0300000000000004e-05,
+      "loss": 0.9439,
+      "step": 403
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0400000000000006e-05,
+      "loss": 0.7106,
+      "step": 404
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.05e-05,
+      "loss": 0.816,
+      "step": 405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0600000000000004e-05,
+      "loss": 0.6328,
+      "step": 406
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.07e-05,
+      "loss": 0.6714,
+      "step": 407
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.08e-05,
+      "loss": 0.7475,
+      "step": 408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.09e-05,
+      "loss": 0.7011,
+      "step": 409
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1e-05,
+      "loss": 0.8192,
+      "step": 410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.11e-05,
+      "loss": 0.8086,
+      "step": 411
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.12e-05,
+      "loss": 0.9789,
+      "step": 412
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.13e-05,
+      "loss": 0.7323,
+      "step": 413
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.14e-05,
+      "loss": 0.9589,
+      "step": 414
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.15e-05,
+      "loss": 0.7151,
+      "step": 415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.16e-05,
+      "loss": 0.7614,
+      "step": 416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.17e-05,
+      "loss": 0.7078,
+      "step": 417
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.18e-05,
+      "loss": 0.8508,
+      "step": 418
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.19e-05,
+      "loss": 0.7595,
+      "step": 419
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2e-05,
+      "loss": 0.629,
+      "step": 420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.21e-05,
+      "loss": 0.6449,
+      "step": 421
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.22e-05,
+      "loss": 0.6758,
+      "step": 422
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.23e-05,
+      "loss": 0.6243,
+      "step": 423
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.24e-05,
+      "loss": 0.9303,
+      "step": 424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.25e-05,
+      "loss": 0.783,
+      "step": 425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.26e-05,
+      "loss": 0.6436,
+      "step": 426
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.27e-05,
+      "loss": 0.671,
+      "step": 427
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2800000000000004e-05,
+      "loss": 0.656,
+      "step": 428
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.29e-05,
+      "loss": 0.7484,
+      "step": 429
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3e-05,
+      "loss": 0.6174,
+      "step": 430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3100000000000004e-05,
+      "loss": 0.5972,
+      "step": 431
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.32e-05,
+      "loss": 0.6567,
+      "step": 432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.33e-05,
+      "loss": 0.7206,
+      "step": 433
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3400000000000005e-05,
+      "loss": 0.715,
+      "step": 434
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.35e-05,
+      "loss": 0.5591,
+      "step": 435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.36e-05,
+      "loss": 0.8661,
+      "step": 436
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3700000000000005e-05,
+      "loss": 0.7022,
+      "step": 437
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.38e-05,
+      "loss": 0.6737,
+      "step": 438
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.39e-05,
+      "loss": 0.625,
+      "step": 439
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.6795,
+      "step": 440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.41e-05,
+      "loss": 0.9394,
+      "step": 441
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4200000000000004e-05,
+      "loss": 0.7992,
+      "step": 442
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.43e-05,
+      "loss": 0.804,
+      "step": 443
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.44e-05,
+      "loss": 0.6675,
+      "step": 444
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4500000000000004e-05,
+      "loss": 0.6051,
+      "step": 445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.46e-05,
+      "loss": 0.8206,
+      "step": 446
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.47e-05,
+      "loss": 0.8177,
+      "step": 447
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4800000000000005e-05,
+      "loss": 0.7704,
+      "step": 448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.49e-05,
+      "loss": 0.6823,
+      "step": 449
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5e-05,
+      "loss": 0.6728,
+      "step": 450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5100000000000005e-05,
+      "loss": 0.6538,
+      "step": 451
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.52e-05,
+      "loss": 0.9582,
+      "step": 452
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.53e-05,
+      "loss": 0.7253,
+      "step": 453
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5400000000000006e-05,
+      "loss": 0.5603,
+      "step": 454
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.55e-05,
+      "loss": 0.6986,
+      "step": 455
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5600000000000004e-05,
+      "loss": 0.5951,
+      "step": 456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5700000000000006e-05,
+      "loss": 0.7598,
+      "step": 457
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.58e-05,
+      "loss": 0.681,
+      "step": 458
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5900000000000004e-05,
+      "loss": 0.7244,
+      "step": 459
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 0.7116,
+      "step": 460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.61e-05,
+      "loss": 0.8005,
+      "step": 461
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6200000000000005e-05,
+      "loss": 0.823,
+      "step": 462
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.630000000000001e-05,
+      "loss": 0.747,
+      "step": 463
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.64e-05,
+      "loss": 0.6731,
+      "step": 464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6500000000000005e-05,
+      "loss": 0.7396,
+      "step": 465
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.660000000000001e-05,
+      "loss": 0.8446,
+      "step": 466
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6700000000000003e-05,
+      "loss": 0.7407,
+      "step": 467
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6800000000000006e-05,
+      "loss": 0.701,
+      "step": 468
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.69e-05,
+      "loss": 0.69,
+      "step": 469
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7e-05,
+      "loss": 0.7501,
+      "step": 470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.71e-05,
+      "loss": 0.72,
+      "step": 471
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.72e-05,
+      "loss": 0.7544,
+      "step": 472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.73e-05,
+      "loss": 0.6247,
+      "step": 473
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.74e-05,
+      "loss": 0.6199,
+      "step": 474
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.75e-05,
+      "loss": 0.6549,
+      "step": 475
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.76e-05,
+      "loss": 0.7425,
+      "step": 476
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.77e-05,
+      "loss": 0.7016,
+      "step": 477
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.78e-05,
+      "loss": 0.7056,
+      "step": 478
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.79e-05,
+      "loss": 0.5853,
+      "step": 479
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8e-05,
+      "loss": 0.6732,
+      "step": 480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8100000000000004e-05,
+      "loss": 0.6233,
+      "step": 481
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.82e-05,
+      "loss": 0.8197,
+      "step": 482
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.83e-05,
+      "loss": 0.7697,
+      "step": 483
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8400000000000004e-05,
+      "loss": 0.7177,
+      "step": 484
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.85e-05,
+      "loss": 0.724,
+      "step": 485
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.86e-05,
+      "loss": 0.7057,
+      "step": 486
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.87e-05,
+      "loss": 0.6493,
+      "step": 487
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.88e-05,
+      "loss": 0.604,
+      "step": 488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.89e-05,
+      "loss": 0.7114,
+      "step": 489
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9e-05,
+      "loss": 0.7439,
+      "step": 490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.91e-05,
+      "loss": 0.7247,
+      "step": 491
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.92e-05,
+      "loss": 0.6159,
+      "step": 492
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.93e-05,
+      "loss": 0.6307,
+      "step": 493
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.94e-05,
+      "loss": 0.616,
+      "step": 494
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9500000000000004e-05,
+      "loss": 0.7276,
+      "step": 495
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.96e-05,
+      "loss": 0.6459,
+      "step": 496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.97e-05,
+      "loss": 0.7552,
+      "step": 497
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9800000000000004e-05,
+      "loss": 0.7142,
+      "step": 498
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99e-05,
+      "loss": 0.7312,
+      "step": 499
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-05,
+      "loss": 0.7048,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999999985823608e-05,
+      "loss": 0.5591,
+      "step": 501
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999999943294431e-05,
+      "loss": 0.6265,
+      "step": 502
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9999998724124696e-05,
+      "loss": 0.6539,
+      "step": 503
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999999773177725e-05,
+      "loss": 0.7367,
+      "step": 504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9999996455901984e-05,
+      "loss": 0.7164,
+      "step": 505
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9999994896498905e-05,
+      "loss": 0.6236,
+      "step": 506
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999999305356804e-05,
+      "loss": 0.6639,
+      "step": 507
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999999092710941e-05,
+      "loss": 0.654,
+      "step": 508
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999998851712302e-05,
+      "loss": 0.5984,
+      "step": 509
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999998582360892e-05,
+      "loss": 0.5919,
+      "step": 510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999998284656714e-05,
+      "loss": 0.7105,
+      "step": 511
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99999795859977e-05,
+      "loss": 0.498,
+      "step": 512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999997604190064e-05,
+      "loss": 0.7987,
+      "step": 513
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999997221427601e-05,
+      "loss": 0.669,
+      "step": 514
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9999968103123844e-05,
+      "loss": 0.7028,
+      "step": 515
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9999963708444196e-05,
+      "loss": 0.5583,
+      "step": 516
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999995903023711e-05,
+      "loss": 0.9376,
+      "step": 517
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999995406850263e-05,
+      "loss": 0.8258,
+      "step": 518
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9999948823240835e-05,
+      "loss": 0.7053,
+      "step": 519
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9999943294451766e-05,
+      "loss": 0.6357,
+      "step": 520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9999937482135494e-05,
+      "loss": 0.789,
+      "step": 521
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9999931386292086e-05,
+      "loss": 0.7267,
+      "step": 522
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999992500692161e-05,
+      "loss": 0.6753,
+      "step": 523
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999991834402413e-05,
+      "loss": 0.702,
+      "step": 524
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999991139759973e-05,
+      "loss": 0.6847,
+      "step": 525
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999990416764848e-05,
+      "loss": 0.6996,
+      "step": 526
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999989665417048e-05,
+      "loss": 0.6223,
+      "step": 527
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99998888571658e-05,
+      "loss": 0.69,
+      "step": 528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9999880776634535e-05,
+      "loss": 0.7145,
+      "step": 529
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9999872412576764e-05,
+      "loss": 0.7823,
+      "step": 530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99998637649926e-05,
+      "loss": 0.6297,
+      "step": 531
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9999854833882134e-05,
+      "loss": 0.7614,
+      "step": 532
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9999845619245464e-05,
+      "loss": 0.7891,
+      "step": 533
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9999836121082696e-05,
+      "loss": 0.7239,
+      "step": 534
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999982633939394e-05,
+      "loss": 0.7175,
+      "step": 535
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9999816274179306e-05,
+      "loss": 0.7226,
+      "step": 536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99998059254389e-05,
+      "loss": 0.7161,
+      "step": 537
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9999795293172857e-05,
+      "loss": 0.6316,
+      "step": 538
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9999784377381284e-05,
+      "loss": 0.726,
+      "step": 539
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999977317806431e-05,
+      "loss": 0.7488,
+      "step": 540
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999976169522206e-05,
+      "loss": 0.585,
+      "step": 541
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999974992885466e-05,
+      "loss": 0.6347,
+      "step": 542
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999973787896225e-05,
+      "loss": 0.5727,
+      "step": 543
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999972554554497e-05,
+      "loss": 0.6969,
+      "step": 544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9999712928602946e-05,
+      "loss": 0.8346,
+      "step": 545
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9999700028136336e-05,
+      "loss": 0.6533,
+      "step": 546
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999968684414528e-05,
+      "loss": 0.5963,
+      "step": 547
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999967337662992e-05,
+      "loss": 0.6459,
+      "step": 548
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999965962559042e-05,
+      "loss": 0.6501,
+      "step": 549
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999964559102694e-05,
+      "loss": 0.6826,
+      "step": 550
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999963127293962e-05,
+      "loss": 0.6599,
+      "step": 551
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999961667132864e-05,
+      "loss": 0.7359,
+      "step": 552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999960178619416e-05,
+      "loss": 0.5728,
+      "step": 553
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999958661753634e-05,
+      "loss": 0.6131,
+      "step": 554
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9999571165355375e-05,
+      "loss": 0.7074,
+      "step": 555
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999955542965141e-05,
+      "loss": 0.6278,
+      "step": 556
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9999539410424646e-05,
+      "loss": 0.6442,
+      "step": 557
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999952310767526e-05,
+      "loss": 0.7081,
+      "step": 558
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999950652140343e-05,
+      "loss": 0.5201,
+      "step": 559
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999948965160935e-05,
+      "loss": 0.5828,
+      "step": 560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999947249829321e-05,
+      "loss": 0.6005,
+      "step": 561
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99994550614552e-05,
+      "loss": 0.6629,
+      "step": 562
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9999437341095535e-05,
+      "loss": 0.5926,
+      "step": 563
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9999419337214394e-05,
+      "loss": 0.5849,
+      "step": 564
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999940104981199e-05,
+      "loss": 0.5851,
+      "step": 565
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999938247888852e-05,
+      "loss": 0.5093,
+      "step": 566
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999936362444422e-05,
+      "loss": 0.6631,
+      "step": 567
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999934448647929e-05,
+      "loss": 0.5989,
+      "step": 568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999932506499394e-05,
+      "loss": 0.714,
+      "step": 569
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99993053599884e-05,
+      "loss": 0.6295,
+      "step": 570
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999928537146289e-05,
+      "loss": 0.6293,
+      "step": 571
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999926509941763e-05,
+      "loss": 0.7918,
+      "step": 572
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9999244543852867e-05,
+      "loss": 0.5222,
+      "step": 573
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9999223704768814e-05,
+      "loss": 0.9182,
+      "step": 574
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999920258216572e-05,
+      "loss": 0.5356,
+      "step": 575
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999918117604382e-05,
+      "loss": 0.7047,
+      "step": 576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9999159486403366e-05,
+      "loss": 0.6233,
+      "step": 577
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9999137513244586e-05,
+      "loss": 0.7967,
+      "step": 578
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9999115256567745e-05,
+      "loss": 0.5048,
+      "step": 579
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999909271637309e-05,
+      "loss": 0.5491,
+      "step": 580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999906989266088e-05,
+      "loss": 0.7306,
+      "step": 581
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999904678543136e-05,
+      "loss": 0.7414,
+      "step": 582
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999902339468481e-05,
+      "loss": 0.6445,
+      "step": 583
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999899972042149e-05,
+      "loss": 0.6918,
+      "step": 584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999897576264166e-05,
+      "loss": 0.6041,
+      "step": 585
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99989515213456e-05,
+      "loss": 0.7146,
+      "step": 586
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999892699653358e-05,
+      "loss": 0.6917,
+      "step": 587
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999890218820589e-05,
+      "loss": 0.8607,
+      "step": 588
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9998877096362795e-05,
+      "loss": 0.7043,
+      "step": 589
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999885172100459e-05,
+      "loss": 0.6634,
+      "step": 590
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999882606213155e-05,
+      "loss": 0.8938,
+      "step": 591
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999880011974399e-05,
+      "loss": 0.7875,
+      "step": 592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999877389384218e-05,
+      "loss": 0.6306,
+      "step": 593
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999874738442643e-05,
+      "loss": 0.7033,
+      "step": 594
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999872059149704e-05,
+      "loss": 0.6269,
+      "step": 595
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9998693515054315e-05,
+      "loss": 0.6759,
+      "step": 596
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999866615509855e-05,
+      "loss": 0.768,
+      "step": 597
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999863851163007e-05,
+      "loss": 0.6648,
+      "step": 598
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9998610584649184e-05,
+      "loss": 0.7211,
+      "step": 599
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999858237415621e-05,
+      "loss": 0.7403,
+      "step": 600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999855388015146e-05,
+      "loss": 0.6845,
+      "step": 601
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9998525102635254e-05,
+      "loss": 0.7154,
+      "step": 602
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999849604160794e-05,
+      "loss": 0.6365,
+      "step": 603
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9998466697069836e-05,
+      "loss": 0.6475,
+      "step": 604
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9998437069021266e-05,
+      "loss": 0.7126,
+      "step": 605
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999840715746257e-05,
+      "loss": 0.6922,
+      "step": 606
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99983769623941e-05,
+      "loss": 0.7027,
+      "step": 607
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9998346483816185e-05,
+      "loss": 0.5649,
+      "step": 608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9998315721729174e-05,
+      "loss": 0.5598,
+      "step": 609
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999828467613341e-05,
+      "loss": 0.8096,
+      "step": 610
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999825334702926e-05,
+      "loss": 0.7318,
+      "step": 611
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9998221734417074e-05,
+      "loss": 0.6736,
+      "step": 612
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99981898382972e-05,
+      "loss": 0.5504,
+      "step": 613
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9998157658670005e-05,
+      "loss": 0.6934,
+      "step": 614
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999812519553586e-05,
+      "loss": 0.9138,
+      "step": 615
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9998092448895136e-05,
+      "loss": 0.6929,
+      "step": 616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9998059418748186e-05,
+      "loss": 0.5758,
+      "step": 617
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9998026105095405e-05,
+      "loss": 0.6197,
+      "step": 618
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999799250793716e-05,
+      "loss": 0.6687,
+      "step": 619
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999795862727383e-05,
+      "loss": 0.7677,
+      "step": 620
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999792446310581e-05,
+      "loss": 0.6177,
+      "step": 621
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999789001543348e-05,
+      "loss": 0.6686,
+      "step": 622
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9997855284257225e-05,
+      "loss": 0.5514,
+      "step": 623
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999782026957745e-05,
+      "loss": 0.6799,
+      "step": 624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999778497139455e-05,
+      "loss": 0.7392,
+      "step": 625
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9997749389708913e-05,
+      "loss": 0.6865,
+      "step": 626
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9997713524520964e-05,
+      "loss": 0.7433,
+      "step": 627
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9997677375831095e-05,
+      "loss": 0.7562,
+      "step": 628
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999764094363971e-05,
+      "loss": 0.6644,
+      "step": 629
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999760422794724e-05,
+      "loss": 0.7321,
+      "step": 630
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9997567228754095e-05,
+      "loss": 0.6516,
+      "step": 631
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9997529946060683e-05,
+      "loss": 0.6726,
+      "step": 632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9997492379867445e-05,
+      "loss": 0.6015,
+      "step": 633
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999745453017479e-05,
+      "loss": 0.5512,
+      "step": 634
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999741639698317e-05,
+      "loss": 0.675,
+      "step": 635
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999737798029299e-05,
+      "loss": 0.6632,
+      "step": 636
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999733928010469e-05,
+      "loss": 0.5965,
+      "step": 637
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9997300296418736e-05,
+      "loss": 0.606,
+      "step": 638
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999726102923554e-05,
+      "loss": 0.5252,
+      "step": 639
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9997221478555564e-05,
+      "loss": 0.643,
+      "step": 640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9997181644379255e-05,
+      "loss": 0.746,
+      "step": 641
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999714152670706e-05,
+      "loss": 0.7508,
+      "step": 642
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999710112553942e-05,
+      "loss": 0.7035,
+      "step": 643
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999706044087683e-05,
+      "loss": 0.7138,
+      "step": 644
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999701947271972e-05,
+      "loss": 0.7178,
+      "step": 645
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999697822106856e-05,
+      "loss": 0.7921,
+      "step": 646
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9996936685923826e-05,
+      "loss": 0.7591,
+      "step": 647
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999689486728599e-05,
+      "loss": 0.6874,
+      "step": 648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9996852765155515e-05,
+      "loss": 0.632,
+      "step": 649
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999681037953289e-05,
+      "loss": 0.7238,
+      "step": 650
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999676771041859e-05,
+      "loss": 0.71,
+      "step": 651
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99967247578131e-05,
+      "loss": 1.0337,
+      "step": 652
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99966815217169e-05,
+      "loss": 0.6828,
+      "step": 653
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999663800213049e-05,
+      "loss": 0.6619,
+      "step": 654
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999659419905437e-05,
+      "loss": 0.916,
+      "step": 655
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9996550112489016e-05,
+      "loss": 0.6645,
+      "step": 656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999650574243494e-05,
+      "loss": 0.7829,
+      "step": 657
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9996461088892653e-05,
+      "loss": 1.1004,
+      "step": 658
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9996416151862645e-05,
+      "loss": 0.7112,
+      "step": 659
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999637093134544e-05,
+      "loss": 1.3414,
+      "step": 660
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999632542734154e-05,
+      "loss": 0.6767,
+      "step": 661
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999627963985147e-05,
+      "loss": 1.194,
+      "step": 662
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999623356887574e-05,
+      "loss": 0.6966,
+      "step": 663
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999618721441488e-05,
+      "loss": 0.6441,
+      "step": 664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999614057646942e-05,
+      "loss": 0.6311,
+      "step": 665
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999609365503987e-05,
+      "loss": 0.6884,
+      "step": 666
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999604645012678e-05,
+      "loss": 0.6741,
+      "step": 667
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999599896173068e-05,
+      "loss": 1.2425,
+      "step": 668
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999595118985211e-05,
+      "loss": 0.9147,
+      "step": 669
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999590313449161e-05,
+      "loss": 0.7125,
+      "step": 670
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9995854795649714e-05,
+      "loss": 0.93,
+      "step": 671
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999580617332699e-05,
+      "loss": 0.7545,
+      "step": 672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999575726752398e-05,
+      "loss": 0.9724,
+      "step": 673
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999570807824124e-05,
+      "loss": 1.0373,
+      "step": 674
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999565860547932e-05,
+      "loss": 0.7364,
+      "step": 675
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99956088492388e-05,
+      "loss": 1.0491,
+      "step": 676
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999555880952023e-05,
+      "loss": 0.6751,
+      "step": 677
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999550848632417e-05,
+      "loss": 0.7648,
+      "step": 678
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9995457879651205e-05,
+      "loss": 0.7744,
+      "step": 679
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999540698950191e-05,
+      "loss": 0.6682,
+      "step": 680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9995355815876856e-05,
+      "loss": 0.9347,
+      "step": 681
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9995304358776624e-05,
+      "loss": 0.9566,
+      "step": 682
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999525261820179e-05,
+      "loss": 0.5542,
+      "step": 683
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999520059415296e-05,
+      "loss": 0.4814,
+      "step": 684
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999514828663071e-05,
+      "loss": 0.783,
+      "step": 685
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999509569563563e-05,
+      "loss": 0.6236,
+      "step": 686
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999504282116833e-05,
+      "loss": 0.5874,
+      "step": 687
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999498966322939e-05,
+      "loss": 0.9555,
+      "step": 688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999493622181943e-05,
+      "loss": 0.7543,
+      "step": 689
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9994882496939053e-05,
+      "loss": 0.6202,
+      "step": 690
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9994828488588866e-05,
+      "loss": 0.6906,
+      "step": 691
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999477419676948e-05,
+      "loss": 0.7148,
+      "step": 692
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9994719621481514e-05,
+      "loss": 0.7029,
+      "step": 693
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999466476272558e-05,
+      "loss": 0.9146,
+      "step": 694
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9994609620502316e-05,
+      "loss": 0.7235,
+      "step": 695
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9994554194812324e-05,
+      "loss": 0.8853,
+      "step": 696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999449848565625e-05,
+      "loss": 0.6488,
+      "step": 697
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999444249303473e-05,
+      "loss": 0.6472,
+      "step": 698
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9994386216948386e-05,
+      "loss": 0.7699,
+      "step": 699
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999432965739785e-05,
+      "loss": 0.957,
+      "step": 700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9994272814383785e-05,
+      "loss": 0.6767,
+      "step": 701
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999421568790682e-05,
+      "loss": 0.5781,
+      "step": 702
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999415827796761e-05,
+      "loss": 0.5353,
+      "step": 703
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999410058456681e-05,
+      "loss": 0.8319,
+      "step": 704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999404260770505e-05,
+      "loss": 0.6127,
+      "step": 705
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9993984347383024e-05,
+      "loss": 1.1507,
+      "step": 706
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999392580360136e-05,
+      "loss": 1.1932,
+      "step": 707
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9993866976360746e-05,
+      "loss": 0.7075,
+      "step": 708
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999380786566183e-05,
+      "loss": 0.7141,
+      "step": 709
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9993748471505295e-05,
+      "loss": 0.5934,
+      "step": 710
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999368879389182e-05,
+      "loss": 0.6757,
+      "step": 711
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9993628832822064e-05,
+      "loss": 0.633,
+      "step": 712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999356858829672e-05,
+      "loss": 0.6683,
+      "step": 713
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999350806031646e-05,
+      "loss": 0.6405,
+      "step": 714
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999344724888199e-05,
+      "loss": 0.7201,
+      "step": 715
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999338615399398e-05,
+      "loss": 0.61,
+      "step": 716
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999332477565313e-05,
+      "loss": 0.9753,
+      "step": 717
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9993263113860134e-05,
+      "loss": 0.6541,
+      "step": 718
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99932011686157e-05,
+      "loss": 0.6817,
+      "step": 719
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9993138939920516e-05,
+      "loss": 0.6457,
+      "step": 720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99930764277753e-05,
+      "loss": 0.6827,
+      "step": 721
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9993013632180766e-05,
+      "loss": 0.6791,
+      "step": 722
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9992950553137604e-05,
+      "loss": 0.7176,
+      "step": 723
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9992887190646555e-05,
+      "loss": 0.7248,
+      "step": 724
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999282354470831e-05,
+      "loss": 0.6663,
+      "step": 725
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9992759615323616e-05,
+      "loss": 0.6197,
+      "step": 726
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999269540249318e-05,
+      "loss": 0.7058,
+      "step": 727
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999263090621775e-05,
+      "loss": 0.6937,
+      "step": 728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999256612649804e-05,
+      "loss": 0.6644,
+      "step": 729
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999250106333479e-05,
+      "loss": 0.6173,
+      "step": 730
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9992435716728734e-05,
+      "loss": 0.5764,
+      "step": 731
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9992370086680626e-05,
+      "loss": 0.623,
+      "step": 732
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99923041731912e-05,
+      "loss": 0.7681,
+      "step": 733
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99922379762612e-05,
+      "loss": 0.7139,
+      "step": 734
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999217149589138e-05,
+      "loss": 0.6516,
+      "step": 735
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99921047320825e-05,
+      "loss": 0.5751,
+      "step": 736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999203768483531e-05,
+      "loss": 0.6544,
+      "step": 737
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999197035415058e-05,
+      "loss": 0.6063,
+      "step": 738
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999190274002906e-05,
+      "loss": 0.474,
+      "step": 739
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999183484247153e-05,
+      "loss": 0.6274,
+      "step": 740
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999176666147876e-05,
+      "loss": 0.7287,
+      "step": 741
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99916981970515e-05,
+      "loss": 0.7022,
+      "step": 742
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999162944919056e-05,
+      "loss": 0.5594,
+      "step": 743
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999156041789669e-05,
+      "loss": 0.6615,
+      "step": 744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999149110317069e-05,
+      "loss": 0.5753,
+      "step": 745
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999142150501335e-05,
+      "loss": 0.6234,
+      "step": 746
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999135162342544e-05,
+      "loss": 0.8409,
+      "step": 747
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999128145840777e-05,
+      "loss": 0.7139,
+      "step": 748
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999121100996113e-05,
+      "loss": 0.6941,
+      "step": 749
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9991140278086316e-05,
+      "loss": 0.6609,
+      "step": 750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999106926278414e-05,
+      "loss": 0.6012,
+      "step": 751
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999099796405539e-05,
+      "loss": 0.6619,
+      "step": 752
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999092638190089e-05,
+      "loss": 0.6421,
+      "step": 753
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999085451632145e-05,
+      "loss": 0.6054,
+      "step": 754
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999078236731788e-05,
+      "loss": 0.7357,
+      "step": 755
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9990709934891e-05,
+      "loss": 0.5906,
+      "step": 756
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999063721904163e-05,
+      "loss": 0.5928,
+      "step": 757
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9990564219770596e-05,
+      "loss": 0.6522,
+      "step": 758
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9990490937078724e-05,
+      "loss": 0.5219,
+      "step": 759
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9990417370966856e-05,
+      "loss": 0.6126,
+      "step": 760
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999034352143581e-05,
+      "loss": 0.5961,
+      "step": 761
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999026938848644e-05,
+      "loss": 0.8068,
+      "step": 762
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999019497211957e-05,
+      "loss": 0.6058,
+      "step": 763
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999012027233605e-05,
+      "loss": 0.5622,
+      "step": 764
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999004528913673e-05,
+      "loss": 0.9082,
+      "step": 765
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998997002252247e-05,
+      "loss": 0.6393,
+      "step": 766
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9989894472494105e-05,
+      "loss": 0.6186,
+      "step": 767
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99898186390525e-05,
+      "loss": 0.6474,
+      "step": 768
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998974252219851e-05,
+      "loss": 0.6657,
+      "step": 769
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998966612193301e-05,
+      "loss": 0.581,
+      "step": 770
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9989589438256856e-05,
+      "loss": 0.5378,
+      "step": 771
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9989512471170926e-05,
+      "loss": 0.6021,
+      "step": 772
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9989435220676086e-05,
+      "loss": 0.5761,
+      "step": 773
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998935768677321e-05,
+      "loss": 0.6797,
+      "step": 774
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998927986946319e-05,
+      "loss": 0.6259,
+      "step": 775
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9989201768746884e-05,
+      "loss": 0.5953,
+      "step": 776
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998912338462521e-05,
+      "loss": 0.5448,
+      "step": 777
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9989044717099034e-05,
+      "loss": 0.7383,
+      "step": 778
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998896576616926e-05,
+      "loss": 0.6791,
+      "step": 779
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9988886531836775e-05,
+      "loss": 0.5874,
+      "step": 780
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998880701410248e-05,
+      "loss": 0.58,
+      "step": 781
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9988727212967277e-05,
+      "loss": 0.6772,
+      "step": 782
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9988647128432075e-05,
+      "loss": 0.5637,
+      "step": 783
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9988566760497776e-05,
+      "loss": 0.596,
+      "step": 784
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99884861091653e-05,
+      "loss": 0.7413,
+      "step": 785
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998840517443555e-05,
+      "loss": 0.5993,
+      "step": 786
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998832395630946e-05,
+      "loss": 0.5894,
+      "step": 787
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9988242454787926e-05,
+      "loss": 0.6025,
+      "step": 788
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99881606698719e-05,
+      "loss": 0.5778,
+      "step": 789
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9988078601562294e-05,
+      "loss": 0.5383,
+      "step": 790
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9987996249860034e-05,
+      "loss": 0.5412,
+      "step": 791
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998791361476607e-05,
+      "loss": 0.6113,
+      "step": 792
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9987830696281335e-05,
+      "loss": 0.655,
+      "step": 793
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998774749440675e-05,
+      "loss": 0.6599,
+      "step": 794
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998766400914329e-05,
+      "loss": 0.5466,
+      "step": 795
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9987580240491886e-05,
+      "loss": 0.4987,
+      "step": 796
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998749618845348e-05,
+      "loss": 0.5738,
+      "step": 797
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998741185302903e-05,
+      "loss": 0.5972,
+      "step": 798
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9987327234219506e-05,
+      "loss": 0.691,
+      "step": 799
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9987242332025846e-05,
+      "loss": 0.6495,
+      "step": 800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9987157146449026e-05,
+      "loss": 0.9047,
+      "step": 801
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9987071677490015e-05,
+      "loss": 0.6367,
+      "step": 802
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9986985925149775e-05,
+      "loss": 0.6399,
+      "step": 803
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998689988942928e-05,
+      "loss": 0.6054,
+      "step": 804
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998681357032952e-05,
+      "loss": 0.7417,
+      "step": 805
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998672696785144e-05,
+      "loss": 0.657,
+      "step": 806
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998664008199606e-05,
+      "loss": 0.5704,
+      "step": 807
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998655291276434e-05,
+      "loss": 0.7699,
+      "step": 808
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998646546015727e-05,
+      "loss": 0.5127,
+      "step": 809
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9986377724175857e-05,
+      "loss": 0.539,
+      "step": 810
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998628970482109e-05,
+      "loss": 0.8264,
+      "step": 811
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998620140209396e-05,
+      "loss": 0.5094,
+      "step": 812
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998611281599548e-05,
+      "loss": 0.6833,
+      "step": 813
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998602394652664e-05,
+      "loss": 0.5965,
+      "step": 814
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998593479368846e-05,
+      "loss": 0.6436,
+      "step": 815
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998584535748195e-05,
+      "loss": 0.5995,
+      "step": 816
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998575563790812e-05,
+      "loss": 0.5499,
+      "step": 817
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998566563496798e-05,
+      "loss": 0.5485,
+      "step": 818
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998557534866257e-05,
+      "loss": 0.5553,
+      "step": 819
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99854847789929e-05,
+      "loss": 0.5302,
+      "step": 820
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998539392596e-05,
+      "loss": 0.6946,
+      "step": 821
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99853027895649e-05,
+      "loss": 0.5128,
+      "step": 822
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9985211369808636e-05,
+      "loss": 0.5361,
+      "step": 823
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998511966669224e-05,
+      "loss": 0.6813,
+      "step": 824
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9985027680216765e-05,
+      "loss": 0.6707,
+      "step": 825
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998493541038324e-05,
+      "loss": 0.5235,
+      "step": 826
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998484285719271e-05,
+      "loss": 0.5871,
+      "step": 827
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998475002064624e-05,
+      "loss": 0.624,
+      "step": 828
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998465690074487e-05,
+      "loss": 0.6231,
+      "step": 829
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998456349748966e-05,
+      "loss": 0.5112,
+      "step": 830
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998446981088166e-05,
+      "loss": 0.5909,
+      "step": 831
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998437584092196e-05,
+      "loss": 0.5991,
+      "step": 832
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9984281587611594e-05,
+      "loss": 0.6659,
+      "step": 833
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998418705095165e-05,
+      "loss": 0.6262,
+      "step": 834
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998409223094319e-05,
+      "loss": 0.5322,
+      "step": 835
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998399712758729e-05,
+      "loss": 0.6323,
+      "step": 836
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998390174088504e-05,
+      "loss": 0.5969,
+      "step": 837
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998380607083751e-05,
+      "loss": 0.6534,
+      "step": 838
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9983710117445795e-05,
+      "loss": 0.6697,
+      "step": 839
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9983613880710966e-05,
+      "loss": 0.7278,
+      "step": 840
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998351736063414e-05,
+      "loss": 0.7051,
+      "step": 841
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998342055721639e-05,
+      "loss": 0.5707,
+      "step": 842
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998332347045882e-05,
+      "loss": 0.5245,
+      "step": 843
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998322610036253e-05,
+      "loss": 0.6024,
+      "step": 844
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998312844692863e-05,
+      "loss": 0.6467,
+      "step": 845
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998303051015822e-05,
+      "loss": 0.6319,
+      "step": 846
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9982932290052426e-05,
+      "loss": 0.7441,
+      "step": 847
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998283378661235e-05,
+      "loss": 0.7805,
+      "step": 848
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99827349998391e-05,
+      "loss": 0.6196,
+      "step": 849
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998263592973381e-05,
+      "loss": 0.5612,
+      "step": 850
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99825365762976e-05,
+      "loss": 0.5329,
+      "step": 851
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99824369395316e-05,
+      "loss": 0.6384,
+      "step": 852
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998233701943693e-05,
+      "loss": 0.6869,
+      "step": 853
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998223681601473e-05,
+      "loss": 0.4618,
+      "step": 854
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9982136329266144e-05,
+      "loss": 0.6635,
+      "step": 855
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99820355591923e-05,
+      "loss": 0.7083,
+      "step": 856
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998193450579434e-05,
+      "loss": 0.6444,
+      "step": 857
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998183316907342e-05,
+      "loss": 0.8772,
+      "step": 858
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9981731549030684e-05,
+      "loss": 0.7544,
+      "step": 859
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9981629645667285e-05,
+      "loss": 0.5697,
+      "step": 860
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9981527458984375e-05,
+      "loss": 0.7847,
+      "step": 861
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9981424988983114e-05,
+      "loss": 0.5158,
+      "step": 862
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998132223566467e-05,
+      "loss": 0.8632,
+      "step": 863
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9981219199030205e-05,
+      "loss": 0.662,
+      "step": 864
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998111587908089e-05,
+      "loss": 0.6467,
+      "step": 865
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998101227581789e-05,
+      "loss": 0.5547,
+      "step": 866
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9980908389242386e-05,
+      "loss": 0.7268,
+      "step": 867
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9980804219355545e-05,
+      "loss": 0.6733,
+      "step": 868
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9980699766158565e-05,
+      "loss": 0.6884,
+      "step": 869
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9980595029652625e-05,
+      "loss": 0.5897,
+      "step": 870
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998049000983891e-05,
+      "loss": 0.6077,
+      "step": 871
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998038470671861e-05,
+      "loss": 0.7488,
+      "step": 872
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9980279120292925e-05,
+      "loss": 0.6862,
+      "step": 873
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9980173250563045e-05,
+      "loss": 0.6949,
+      "step": 874
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998006709753017e-05,
+      "loss": 0.576,
+      "step": 875
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9979960661195514e-05,
+      "loss": 0.5039,
+      "step": 876
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997985394156028e-05,
+      "loss": 0.6543,
+      "step": 877
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9979746938625674e-05,
+      "loss": 0.5377,
+      "step": 878
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997963965239291e-05,
+      "loss": 0.5142,
+      "step": 879
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997953208286321e-05,
+      "loss": 0.593,
+      "step": 880
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9979424230037794e-05,
+      "loss": 0.6123,
+      "step": 881
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997931609391787e-05,
+      "loss": 0.5768,
+      "step": 882
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997920767450469e-05,
+      "loss": 0.6576,
+      "step": 883
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997909897179946e-05,
+      "loss": 0.5991,
+      "step": 884
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997898998580343e-05,
+      "loss": 0.6187,
+      "step": 885
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997888071651783e-05,
+      "loss": 0.6767,
+      "step": 886
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9978771163943895e-05,
+      "loss": 0.4948,
+      "step": 887
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997866132808287e-05,
+      "loss": 0.4886,
+      "step": 888
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9978551208936e-05,
+      "loss": 0.9106,
+      "step": 889
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9978440806504533e-05,
+      "loss": 0.7495,
+      "step": 890
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997833012078973e-05,
+      "loss": 0.5169,
+      "step": 891
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997821915179284e-05,
+      "loss": 0.5922,
+      "step": 892
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9978107899515116e-05,
+      "loss": 0.6284,
+      "step": 893
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9977996363957824e-05,
+      "loss": 0.612,
+      "step": 894
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9977884545122233e-05,
+      "loss": 0.5493,
+      "step": 895
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9977772443009607e-05,
+      "loss": 0.7183,
+      "step": 896
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997766005762121e-05,
+      "loss": 0.5678,
+      "step": 897
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997754738895833e-05,
+      "loss": 0.7236,
+      "step": 898
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9977434437022245e-05,
+      "loss": 0.5603,
+      "step": 899
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997732120181423e-05,
+      "loss": 0.5691,
+      "step": 900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997720768333557e-05,
+      "loss": 0.4718,
+      "step": 901
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997709388158754e-05,
+      "loss": 0.5686,
+      "step": 902
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9976979796571456e-05,
+      "loss": 0.5364,
+      "step": 903
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99768654282886e-05,
+      "loss": 0.4634,
+      "step": 904
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9976750776740253e-05,
+      "loss": 0.5733,
+      "step": 905
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9976635841927745e-05,
+      "loss": 0.6549,
+      "step": 906
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997652062385236e-05,
+      "loss": 0.6736,
+      "step": 907
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9976405122515406e-05,
+      "loss": 0.5853,
+      "step": 908
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9976289337918205e-05,
+      "loss": 0.5687,
+      "step": 909
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9976173270062046e-05,
+      "loss": 0.5265,
+      "step": 910
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997605691894827e-05,
+      "loss": 0.7175,
+      "step": 911
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997594028457819e-05,
+      "loss": 0.4941,
+      "step": 912
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9975823366953124e-05,
+      "loss": 0.5198,
+      "step": 913
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9975706166074395e-05,
+      "loss": 0.5668,
+      "step": 914
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997558868194334e-05,
+      "loss": 0.5769,
+      "step": 915
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997547091456128e-05,
+      "loss": 0.5132,
+      "step": 916
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997535286392958e-05,
+      "loss": 0.7157,
+      "step": 917
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9975234530049544e-05,
+      "loss": 0.6017,
+      "step": 918
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9975115912922525e-05,
+      "loss": 0.6287,
+      "step": 919
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997499701254987e-05,
+      "loss": 0.613,
+      "step": 920
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997487782893294e-05,
+      "loss": 0.6579,
+      "step": 921
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997475836207307e-05,
+      "loss": 0.6522,
+      "step": 922
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997463861197162e-05,
+      "loss": 0.5591,
+      "step": 923
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997451857862995e-05,
+      "loss": 0.5891,
+      "step": 924
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997439826204942e-05,
+      "loss": 0.559,
+      "step": 925
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997427766223138e-05,
+      "loss": 0.5309,
+      "step": 926
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9974156779177226e-05,
+      "loss": 0.6118,
+      "step": 927
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9974035612888316e-05,
+      "loss": 0.5986,
+      "step": 928
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997391416336602e-05,
+      "loss": 0.7495,
+      "step": 929
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997379243061172e-05,
+      "loss": 0.6227,
+      "step": 930
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997367041462679e-05,
+      "loss": 0.6105,
+      "step": 931
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997354811541262e-05,
+      "loss": 0.6013,
+      "step": 932
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99734255329706e-05,
+      "loss": 0.5306,
+      "step": 933
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9973302667302114e-05,
+      "loss": 0.5988,
+      "step": 934
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997317951840855e-05,
+      "loss": 0.5232,
+      "step": 935
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997305608629132e-05,
+      "loss": 0.9796,
+      "step": 936
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9972932370951816e-05,
+      "loss": 0.5463,
+      "step": 937
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997280837239144e-05,
+      "loss": 0.657,
+      "step": 938
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99726840906116e-05,
+      "loss": 0.5983,
+      "step": 939
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99725595256137e-05,
+      "loss": 0.574,
+      "step": 940
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9972434677399166e-05,
+      "loss": 0.6257,
+      "step": 941
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99723095459694e-05,
+      "loss": 0.6569,
+      "step": 942
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997218413132583e-05,
+      "loss": 0.5688,
+      "step": 943
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997205843346987e-05,
+      "loss": 0.5806,
+      "step": 944
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9971932452402956e-05,
+      "loss": 0.5838,
+      "step": 945
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99718061881265e-05,
+      "loss": 0.5894,
+      "step": 946
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9971679640641956e-05,
+      "loss": 0.6019,
+      "step": 947
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997155280995074e-05,
+      "loss": 0.5362,
+      "step": 948
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997142569605431e-05,
+      "loss": 0.457,
+      "step": 949
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997129829895409e-05,
+      "loss": 0.5174,
+      "step": 950
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997117061865153e-05,
+      "loss": 0.5131,
+      "step": 951
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9971042655148083e-05,
+      "loss": 0.5786,
+      "step": 952
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9970914408445194e-05,
+      "loss": 0.6363,
+      "step": 953
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997078587854431e-05,
+      "loss": 0.5756,
+      "step": 954
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997065706544691e-05,
+      "loss": 0.5431,
+      "step": 955
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9970527969154443e-05,
+      "loss": 0.5649,
+      "step": 956
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9970398589668376e-05,
+      "loss": 0.5451,
+      "step": 957
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997026892699017e-05,
+      "loss": 0.5719,
+      "step": 958
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99701389811213e-05,
+      "loss": 0.5627,
+      "step": 959
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997000875206323e-05,
+      "loss": 0.6205,
+      "step": 960
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996987823981746e-05,
+      "loss": 0.591,
+      "step": 961
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9969747444385453e-05,
+      "loss": 0.5994,
+      "step": 962
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9969616365768695e-05,
+      "loss": 0.4875,
+      "step": 963
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996948500396868e-05,
+      "loss": 0.6434,
+      "step": 964
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9969353358986876e-05,
+      "loss": 0.5786,
+      "step": 965
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9969221430824796e-05,
+      "loss": 0.4642,
+      "step": 966
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9969089219483935e-05,
+      "loss": 0.6638,
+      "step": 967
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9968956724965784e-05,
+      "loss": 0.721,
+      "step": 968
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996882394727186e-05,
+      "loss": 0.5264,
+      "step": 969
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996869088640365e-05,
+      "loss": 0.5226,
+      "step": 970
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996855754236267e-05,
+      "loss": 0.5806,
+      "step": 971
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996842391515044e-05,
+      "loss": 0.6467,
+      "step": 972
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996829000476847e-05,
+      "loss": 0.5,
+      "step": 973
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996815581121828e-05,
+      "loss": 0.5105,
+      "step": 974
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996802133450138e-05,
+      "loss": 0.6887,
+      "step": 975
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996788657461931e-05,
+      "loss": 0.6605,
+      "step": 976
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9967751531573604e-05,
+      "loss": 0.4807,
+      "step": 977
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996761620536578e-05,
+      "loss": 0.5663,
+      "step": 978
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996748059599737e-05,
+      "loss": 0.553,
+      "step": 979
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996734470346992e-05,
+      "loss": 0.6024,
+      "step": 980
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996720852778497e-05,
+      "loss": 0.4422,
+      "step": 981
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9967072068944065e-05,
+      "loss": 0.4934,
+      "step": 982
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996693532694875e-05,
+      "loss": 0.3896,
+      "step": 983
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9966798301800575e-05,
+      "loss": 0.6925,
+      "step": 984
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99666609935011e-05,
+      "loss": 0.5968,
+      "step": 985
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996652340205188e-05,
+      "loss": 0.606,
+      "step": 986
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9966385527454466e-05,
+      "loss": 0.5086,
+      "step": 987
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996624736971044e-05,
+      "loss": 0.5204,
+      "step": 988
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996610892882135e-05,
+      "loss": 0.355,
+      "step": 989
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9965970204788775e-05,
+      "loss": 0.4276,
+      "step": 990
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996583119761429e-05,
+      "loss": 0.7765,
+      "step": 991
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9965691907299474e-05,
+      "loss": 0.8122,
+      "step": 992
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996555233384589e-05,
+      "loss": 0.5152,
+      "step": 993
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9965412477255145e-05,
+      "loss": 0.5286,
+      "step": 994
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9965272337528803e-05,
+      "loss": 0.5542,
+      "step": 995
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996513191466847e-05,
+      "loss": 0.5672,
+      "step": 996
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9964991208675724e-05,
+      "loss": 0.4148,
+      "step": 997
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9964850219552176e-05,
+      "loss": 0.4722,
+      "step": 998
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9964708947299424e-05,
+      "loss": 0.6233,
+      "step": 999
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996456739191905e-05,
+      "loss": 0.7077,
+      "step": 1000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996442555341268e-05,
+      "loss": 0.4757,
+      "step": 1001
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996428343178191e-05,
+      "loss": 0.4317,
+      "step": 1002
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9964141027028364e-05,
+      "loss": 0.7334,
+      "step": 1003
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996399833915365e-05,
+      "loss": 0.4498,
+      "step": 1004
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996385536815938e-05,
+      "loss": 0.7786,
+      "step": 1005
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9963712114047196e-05,
+      "loss": 1.4099,
+      "step": 1006
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99635685768187e-05,
+      "loss": 0.4422,
+      "step": 1007
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996342475647553e-05,
+      "loss": 0.5521,
+      "step": 1008
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996328065301932e-05,
+      "loss": 0.6481,
+      "step": 1009
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996313626645169e-05,
+      "loss": 0.6351,
+      "step": 1010
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9962991596774296e-05,
+      "loss": 0.4226,
+      "step": 1011
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996284664398877e-05,
+      "loss": 0.5085,
+      "step": 1012
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996270140809675e-05,
+      "loss": 0.5383,
+      "step": 1013
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9962555889099895e-05,
+      "loss": 0.4848,
+      "step": 1014
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996241008699985e-05,
+      "loss": 0.5958,
+      "step": 1015
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996226400179827e-05,
+      "loss": 0.4303,
+      "step": 1016
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996211763349681e-05,
+      "loss": 0.4896,
+      "step": 1017
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996197098209712e-05,
+      "loss": 0.4847,
+      "step": 1018
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996182404760088e-05,
+      "loss": 0.4906,
+      "step": 1019
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9961676830009754e-05,
+      "loss": 0.4529,
+      "step": 1020
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9961529329325406e-05,
+      "loss": 0.5284,
+      "step": 1021
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996138154554951e-05,
+      "loss": 0.5634,
+      "step": 1022
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996123347868373e-05,
+      "loss": 0.5425,
+      "step": 1023
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9961085128729774e-05,
+      "loss": 0.5843,
+      "step": 1024
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99609364956893e-05,
+      "loss": 0.4616,
+      "step": 1025
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996078757956401e-05,
+      "loss": 0.4211,
+      "step": 1026
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996063838035557e-05,
+      "loss": 0.5223,
+      "step": 1027
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996048889806569e-05,
+      "loss": 0.5464,
+      "step": 1028
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9960339132696077e-05,
+      "loss": 1.2438,
+      "step": 1029
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99601890842484e-05,
+      "loss": 1.5612,
+      "step": 1030
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996003875272438e-05,
+      "loss": 1.4655,
+      "step": 1031
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995988813812571e-05,
+      "loss": 2.9384,
+      "step": 1032
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995973724045412e-05,
+      "loss": 0.9683,
+      "step": 1033
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995958605971129e-05,
+      "loss": 1.6349,
+      "step": 1034
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9959434595898955e-05,
+      "loss": 2.1343,
+      "step": 1035
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995928284901884e-05,
+      "loss": 1.4716,
+      "step": 1036
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995913081907264e-05,
+      "loss": 1.1104,
+      "step": 1037
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9958978506062105e-05,
+      "loss": 1.867,
+      "step": 1038
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995882590998895e-05,
+      "loss": 1.6188,
+      "step": 1039
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9958673030854896e-05,
+      "loss": 1.5363,
+      "step": 1040
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99585198686617e-05,
+      "loss": 2.0023,
+      "step": 1041
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995836642341108e-05,
+      "loss": 1.5455,
+      "step": 1042
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9958212695104786e-05,
+      "loss": 1.8537,
+      "step": 1043
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9958058683744555e-05,
+      "loss": 1.2362,
+      "step": 1044
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9957904389332144e-05,
+      "loss": 2.0989,
+      "step": 1045
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9957749811869294e-05,
+      "loss": 1.5746,
+      "step": 1046
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9957594951357755e-05,
+      "loss": 1.2454,
+      "step": 1047
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995743980779929e-05,
+      "loss": 1.5064,
+      "step": 1048
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9957284381195665e-05,
+      "loss": 2.4676,
+      "step": 1049
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9957128671548625e-05,
+      "loss": 1.2908,
+      "step": 1050
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995697267885995e-05,
+      "loss": 1.1533,
+      "step": 1051
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99568164031314e-05,
+      "loss": 1.1945,
+      "step": 1052
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995665984436476e-05,
+      "loss": 1.1818,
+      "step": 1053
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99565030025618e-05,
+      "loss": 1.1864,
+      "step": 1054
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995634587772428e-05,
+      "loss": 1.4476,
+      "step": 1055
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995618846985401e-05,
+      "loss": 1.4678,
+      "step": 1056
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995603077895276e-05,
+      "loss": 1.3017,
+      "step": 1057
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995587280502232e-05,
+      "loss": 1.3511,
+      "step": 1058
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9955714548064495e-05,
+      "loss": 1.174,
+      "step": 1059
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995555600808105e-05,
+      "loss": 0.9336,
+      "step": 1060
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995539718507382e-05,
+      "loss": 0.8242,
+      "step": 1061
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9955238079044576e-05,
+      "loss": 1.776,
+      "step": 1062
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995507868999513e-05,
+      "loss": 2.1563,
+      "step": 1063
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995491901792731e-05,
+      "loss": 2.1084,
+      "step": 1064
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995475906284289e-05,
+      "loss": 1.4557,
+      "step": 1065
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995459882474372e-05,
+      "loss": 0.7916,
+      "step": 1066
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995443830363159e-05,
+      "loss": 1.154,
+      "step": 1067
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995427749950835e-05,
+      "loss": 1.3613,
+      "step": 1068
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995411641237579e-05,
+      "loss": 2.0577,
+      "step": 1069
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995395504223575e-05,
+      "loss": 2.1272,
+      "step": 1070
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9953793389090083e-05,
+      "loss": 1.1423,
+      "step": 1071
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995363145294058e-05,
+      "loss": 1.7777,
+      "step": 1072
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995346923378912e-05,
+      "loss": 2.0965,
+      "step": 1073
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995330673163751e-05,
+      "loss": 2.2885,
+      "step": 1074
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995314394648761e-05,
+      "loss": 0.8863,
+      "step": 1075
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995298087834126e-05,
+      "loss": 1.9341,
+      "step": 1076
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9952817527200313e-05,
+      "loss": 2.3828,
+      "step": 1077
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995265389306661e-05,
+      "loss": 1.0734,
+      "step": 1078
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995248997594203e-05,
+      "loss": 1.1712,
+      "step": 1079
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995232577582841e-05,
+      "loss": 0.8423,
+      "step": 1080
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9952161292727627e-05,
+      "loss": 1.5221,
+      "step": 1081
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9951996526641544e-05,
+      "loss": 1.17,
+      "step": 1082
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995183147757201e-05,
+      "loss": 1.0132,
+      "step": 1083
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995166614552092e-05,
+      "loss": 1.748,
+      "step": 1084
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9951500530490146e-05,
+      "loss": 1.5261,
+      "step": 1085
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995133463248156e-05,
+      "loss": 2.187,
+      "step": 1086
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9951168451497044e-05,
+      "loss": 1.1532,
+      "step": 1087
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995100198753849e-05,
+      "loss": 0.7479,
+      "step": 1088
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995083524060777e-05,
+      "loss": 1.4198,
+      "step": 1089
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995066821070679e-05,
+      "loss": 0.8311,
+      "step": 1090
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995050089783744e-05,
+      "loss": 1.1347,
+      "step": 1091
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995033330200162e-05,
+      "loss": 2.106,
+      "step": 1092
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995016542320122e-05,
+      "loss": 1.4183,
+      "step": 1093
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9949997261438155e-05,
+      "loss": 0.8613,
+      "step": 1094
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994982881671433e-05,
+      "loss": 1.4838,
+      "step": 1095
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994966008903165e-05,
+      "loss": 2.3552,
+      "step": 1096
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9949491078392034e-05,
+      "loss": 1.1441,
+      "step": 1097
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9949321784797395e-05,
+      "loss": 1.4423,
+      "step": 1098
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994915220824966e-05,
+      "loss": 1.5135,
+      "step": 1099
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994898234875075e-05,
+      "loss": 1.7208,
+      "step": 1100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9948812206302584e-05,
+      "loss": 1.4972,
+      "step": 1101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99486417809071e-05,
+      "loss": 0.4514,
+      "step": 1102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9948471072566226e-05,
+      "loss": 0.5349,
+      "step": 1103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9948300081281896e-05,
+      "loss": 0.5492,
+      "step": 1104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994812880705606e-05,
+      "loss": 0.4701,
+      "step": 1105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994795724989064e-05,
+      "loss": 0.4489,
+      "step": 1106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99477854097876e-05,
+      "loss": 0.5719,
+      "step": 1107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9947613286748894e-05,
+      "loss": 0.6232,
+      "step": 1108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994744088077646e-05,
+      "loss": 0.513,
+      "step": 1109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9947268191872256e-05,
+      "loss": 0.4267,
+      "step": 1110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994709522003823e-05,
+      "loss": 0.4808,
+      "step": 1111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994692196527637e-05,
+      "loss": 0.617,
+      "step": 1112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994674842758862e-05,
+      "loss": 0.4874,
+      "step": 1113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994657460697695e-05,
+      "loss": 0.621,
+      "step": 1114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9946400503443336e-05,
+      "loss": 0.4756,
+      "step": 1115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9946226116989755e-05,
+      "loss": 0.5579,
+      "step": 1116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994605144761818e-05,
+      "loss": 0.4585,
+      "step": 1117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99458764953306e-05,
+      "loss": 0.629,
+      "step": 1118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994570126012898e-05,
+      "loss": 0.6288,
+      "step": 1119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994552574201533e-05,
+      "loss": 0.5068,
+      "step": 1120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9945349940991634e-05,
+      "loss": 0.5252,
+      "step": 1121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9945173857059876e-05,
+      "loss": 0.5787,
+      "step": 1122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9944997490222055e-05,
+      "loss": 0.5247,
+      "step": 1123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994482084048018e-05,
+      "loss": 0.3993,
+      "step": 1124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994464390783625e-05,
+      "loss": 0.6135,
+      "step": 1125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994446669229228e-05,
+      "loss": 0.4146,
+      "step": 1126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994428919385026e-05,
+      "loss": 0.4205,
+      "step": 1127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994411141251222e-05,
+      "loss": 0.581,
+      "step": 1128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994393334828017e-05,
+      "loss": 0.6376,
+      "step": 1129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994375500115613e-05,
+      "loss": 0.5085,
+      "step": 1130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994357637114212e-05,
+      "loss": 0.4683,
+      "step": 1131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994339745824017e-05,
+      "loss": 0.6002,
+      "step": 1132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9943218262452314e-05,
+      "loss": 0.4932,
+      "step": 1133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994303878378057e-05,
+      "loss": 0.4162,
+      "step": 1134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9942859022226985e-05,
+      "loss": 0.437,
+      "step": 1135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9942678977793596e-05,
+      "loss": 0.4819,
+      "step": 1136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994249865048244e-05,
+      "loss": 0.51,
+      "step": 1137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994231804029557e-05,
+      "loss": 0.4742,
+      "step": 1138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9942137147235026e-05,
+      "loss": 0.473,
+      "step": 1139
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994195597130286e-05,
+      "loss": 0.4951,
+      "step": 1140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994177451250113e-05,
+      "loss": 0.4712,
+      "step": 1141
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99415927708319e-05,
+      "loss": 0.4613,
+      "step": 1142
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9941410746297226e-05,
+      "loss": 0.6105,
+      "step": 1143
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994122843889916e-05,
+      "loss": 0.8021,
+      "step": 1144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994104584863979e-05,
+      "loss": 0.4298,
+      "step": 1145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9940862975521175e-05,
+      "loss": 0.4732,
+      "step": 1146
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994067981954539e-05,
+      "loss": 0.3977,
+      "step": 1147
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994049638071452e-05,
+      "loss": 0.5213,
+      "step": 1148
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994031265903063e-05,
+      "loss": 0.4872,
+      "step": 1149
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994012865449582e-05,
+      "loss": 0.5332,
+      "step": 1150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993994436711217e-05,
+      "loss": 0.5737,
+      "step": 1151
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993975979688177e-05,
+      "loss": 0.4863,
+      "step": 1152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9939574943806706e-05,
+      "loss": 0.5822,
+      "step": 1153
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993938980788909e-05,
+      "loss": 0.5841,
+      "step": 1154
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9939204389131004e-05,
+      "loss": 0.4514,
+      "step": 1155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993901868753457e-05,
+      "loss": 0.4436,
+      "step": 1156
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9938832703101874e-05,
+      "loss": 0.6127,
+      "step": 1157
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993864643583504e-05,
+      "loss": 0.4352,
+      "step": 1158
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993845988573618e-05,
+      "loss": 0.4043,
+      "step": 1159
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99382730528074e-05,
+      "loss": 0.4082,
+      "step": 1160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993808593705082e-05,
+      "loss": 0.6228,
+      "step": 1161
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993789853846858e-05,
+      "loss": 0.5527,
+      "step": 1162
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9937710857062775e-05,
+      "loss": 0.5784,
+      "step": 1163
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993752289283556e-05,
+      "loss": 0.5128,
+      "step": 1164
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993733464578906e-05,
+      "loss": 0.5203,
+      "step": 1165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9937146115925396e-05,
+      "loss": 0.46,
+      "step": 1166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993695730324672e-05,
+      "loss": 0.4957,
+      "step": 1167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993676820775518e-05,
+      "loss": 0.6604,
+      "step": 1168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99365788294529e-05,
+      "loss": 0.509,
+      "step": 1169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993638916834205e-05,
+      "loss": 0.5012,
+      "step": 1170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9936199224424756e-05,
+      "loss": 0.5102,
+      "step": 1171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9936008997703195e-05,
+      "loss": 0.6358,
+      "step": 1172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993581848817951e-05,
+      "loss": 0.6147,
+      "step": 1173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9935627695855866e-05,
+      "loss": 0.7647,
+      "step": 1174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9935436620734436e-05,
+      "loss": 0.526,
+      "step": 1175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993524526281737e-05,
+      "loss": 0.5493,
+      "step": 1176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993505362210685e-05,
+      "loss": 0.3901,
+      "step": 1177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993486169860505e-05,
+      "loss": 0.3862,
+      "step": 1178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993466949231413e-05,
+      "loss": 0.419,
+      "step": 1179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9934477003236294e-05,
+      "loss": 0.5916,
+      "step": 1180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993428423137372e-05,
+      "loss": 0.4858,
+      "step": 1181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993409117672857e-05,
+      "loss": 0.6019,
+      "step": 1182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9933897839303065e-05,
+      "loss": 0.5352,
+      "step": 1183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993370421909938e-05,
+      "loss": 0.7194,
+      "step": 1184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9933510316119716e-05,
+      "loss": 0.5403,
+      "step": 1185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993331613036627e-05,
+      "loss": 0.5966,
+      "step": 1186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9933121661841244e-05,
+      "loss": 0.6225,
+      "step": 1187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993292691054685e-05,
+      "loss": 0.5953,
+      "step": 1188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993273187648528e-05,
+      "loss": 0.5446,
+      "step": 1189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9932536559658767e-05,
+      "loss": 0.3778,
+      "step": 1190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9932340960069513e-05,
+      "loss": 0.6198,
+      "step": 1191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993214507771974e-05,
+      "loss": 0.5016,
+      "step": 1192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993194891261167e-05,
+      "loss": 0.4696,
+      "step": 1193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993175246474753e-05,
+      "loss": 0.4846,
+      "step": 1194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9931555734129546e-05,
+      "loss": 0.7254,
+      "step": 1195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993135872075994e-05,
+      "loss": 0.6209,
+      "step": 1196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993116142464096e-05,
+      "loss": 0.9184,
+      "step": 1197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993096384577484e-05,
+      "loss": 0.4594,
+      "step": 1198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9930765984163804e-05,
+      "loss": 0.545,
+      "step": 1199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993056783981013e-05,
+      "loss": 0.6107,
+      "step": 1200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993036941271604e-05,
+      "loss": 0.5256,
+      "step": 1201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9930170702883786e-05,
+      "loss": 0.399,
+      "step": 1202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992997171031563e-05,
+      "loss": 0.534,
+      "step": 1203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992977243501382e-05,
+      "loss": 0.5851,
+      "step": 1204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9929572876980624e-05,
+      "loss": 0.5176,
+      "step": 1205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9929373036218305e-05,
+      "loss": 0.5664,
+      "step": 1206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992917291272912e-05,
+      "loss": 0.5316,
+      "step": 1207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992897250651535e-05,
+      "loss": 0.4987,
+      "step": 1208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992877181757926e-05,
+      "loss": 0.5118,
+      "step": 1209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992857084592313e-05,
+      "loss": 0.5314,
+      "step": 1210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992836959154924e-05,
+      "loss": 0.5032,
+      "step": 1211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992816805445987e-05,
+      "loss": 0.4123,
+      "step": 1212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9927966234657306e-05,
+      "loss": 0.4261,
+      "step": 1213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9927764132143836e-05,
+      "loss": 0.4729,
+      "step": 1214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992756174692175e-05,
+      "loss": 0.5285,
+      "step": 1215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992735907899335e-05,
+      "loss": 0.5642,
+      "step": 1216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992715612836093e-05,
+      "loss": 0.4459,
+      "step": 1217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992695289502679e-05,
+      "loss": 0.5769,
+      "step": 1218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992674937899324e-05,
+      "loss": 0.5055,
+      "step": 1219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992654558026259e-05,
+      "loss": 0.453,
+      "step": 1220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992634149883714e-05,
+      "loss": 0.4727,
+      "step": 1221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992613713471922e-05,
+      "loss": 0.8104,
+      "step": 1222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9925932487911134e-05,
+      "loss": 0.3613,
+      "step": 1223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99257275584152e-05,
+      "loss": 0.446,
+      "step": 1224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992552234623376e-05,
+      "loss": 0.5536,
+      "step": 1225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992531685136913e-05,
+      "loss": 0.4312,
+      "step": 1226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9925111073823635e-05,
+      "loss": 0.5304,
+      "step": 1227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992490501359963e-05,
+      "loss": 0.5159,
+      "step": 1228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9924698670699425e-05,
+      "loss": 0.7132,
+      "step": 1229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992449204512538e-05,
+      "loss": 0.4705,
+      "step": 1230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992428513687983e-05,
+      "loss": 0.4711,
+      "step": 1231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992407794596512e-05,
+      "loss": 0.5146,
+      "step": 1232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99238704723836e-05,
+      "loss": 0.4325,
+      "step": 1233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992366271613763e-05,
+      "loss": 0.4828,
+      "step": 1234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9923454677229565e-05,
+      "loss": 0.4494,
+      "step": 1235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992324635566176e-05,
+      "loss": 0.4578,
+      "step": 1236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9923037751436576e-05,
+      "loss": 0.4349,
+      "step": 1237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992282886455638e-05,
+      "loss": 0.4929,
+      "step": 1238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992261969502354e-05,
+      "loss": 0.4562,
+      "step": 1239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992241024284043e-05,
+      "loss": 0.4613,
+      "step": 1240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9922200508009434e-05,
+      "loss": 0.4411,
+      "step": 1241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992199049053292e-05,
+      "loss": 0.5059,
+      "step": 1242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9921780190413277e-05,
+      "loss": 0.4243,
+      "step": 1243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992156960765288e-05,
+      "loss": 0.5276,
+      "step": 1244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992135874225412e-05,
+      "loss": 0.4458,
+      "step": 1245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99211475942194e-05,
+      "loss": 0.4521,
+      "step": 1246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99209361635511e-05,
+      "loss": 0.4579,
+      "step": 1247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992072445025163e-05,
+      "loss": 0.414,
+      "step": 1248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992051245432338e-05,
+      "loss": 0.5579,
+      "step": 1249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992030017576875e-05,
+      "loss": 0.4607,
+      "step": 1250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9920087614590174e-05,
+      "loss": 0.4716,
+      "step": 1251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991987477079004e-05,
+      "loss": 0.4644,
+      "step": 1252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991966164437076e-05,
+      "loss": 0.608,
+      "step": 1253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991944823533476e-05,
+      "loss": 0.4966,
+      "step": 1254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991923454368447e-05,
+      "loss": 0.4054,
+      "step": 1255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991902056942229e-05,
+      "loss": 0.4297,
+      "step": 1256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991880631255066e-05,
+      "loss": 0.6256,
+      "step": 1257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991859177307202e-05,
+      "loss": 0.5,
+      "step": 1258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991837695098878e-05,
+      "loss": 0.5162,
+      "step": 1259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9918161846303394e-05,
+      "loss": 0.5029,
+      "step": 1260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991794645901829e-05,
+      "loss": 0.4914,
+      "step": 1261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991773078913592e-05,
+      "loss": 0.4461,
+      "step": 1262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991751483665873e-05,
+      "loss": 0.4518,
+      "step": 1263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991729860158916e-05,
+      "loss": 0.4448,
+      "step": 1264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991708208392967e-05,
+      "loss": 0.4017,
+      "step": 1265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991686528368271e-05,
+      "loss": 0.4397,
+      "step": 1266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991664820085074e-05,
+      "loss": 0.4493,
+      "step": 1267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991643083543624e-05,
+      "loss": 0.5931,
+      "step": 1268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991621318744165e-05,
+      "loss": 0.5448,
+      "step": 1269
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9915995256869444e-05,
+      "loss": 0.4977,
+      "step": 1270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9915777043722104e-05,
+      "loss": 0.4368,
+      "step": 1271
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991555854800209e-05,
+      "loss": 0.4678,
+      "step": 1272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99153397697119e-05,
+      "loss": 0.3804,
+      "step": 1273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991512070885399e-05,
+      "loss": 0.4156,
+      "step": 1274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9914901365430866e-05,
+      "loss": 0.4237,
+      "step": 1275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9914681739445004e-05,
+      "loss": 0.5098,
+      "step": 1276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99144618308989e-05,
+      "loss": 0.4747,
+      "step": 1277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991424163979504e-05,
+      "loss": 0.5992,
+      "step": 1278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991402116613593e-05,
+      "loss": 0.4387,
+      "step": 1279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991380040992407e-05,
+      "loss": 0.3996,
+      "step": 1280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991357937116196e-05,
+      "loss": 0.493,
+      "step": 1281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991335804985211e-05,
+      "loss": 0.476,
+      "step": 1282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9913136445997026e-05,
+      "loss": 0.4492,
+      "step": 1283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991291455959922e-05,
+      "loss": 0.5053,
+      "step": 1284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991269239066122e-05,
+      "loss": 0.5035,
+      "step": 1285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991246993918553e-05,
+      "loss": 0.3808,
+      "step": 1286
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991224720517469e-05,
+      "loss": 0.3111,
+      "step": 1287
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991202418863121e-05,
+      "loss": 0.4845,
+      "step": 1288
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9911800889557625e-05,
+      "loss": 0.4447,
+      "step": 1289
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991157730795647e-05,
+      "loss": 0.465,
+      "step": 1290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991135344383028e-05,
+      "loss": 0.4888,
+      "step": 1291
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991112929718159e-05,
+      "loss": 0.4036,
+      "step": 1292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9910904868012945e-05,
+      "loss": 0.5542,
+      "step": 1293
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9910680156326885e-05,
+      "loss": 0.5942,
+      "step": 1294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991045516212597e-05,
+      "loss": 0.5431,
+      "step": 1295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.991022988541274e-05,
+      "loss": 0.6387,
+      "step": 1296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9910004326189765e-05,
+      "loss": 0.4931,
+      "step": 1297
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.990977848445959e-05,
+      "loss": 0.5479,
+      "step": 1298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.990955236022477e-05,
+      "loss": 0.5017,
+      "step": 1299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.990932595348789e-05,
+      "loss": 0.4999,
+      "step": 1300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9909099264251495e-05,
+      "loss": 0.5498,
+      "step": 1301
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.990887229251817e-05,
+      "loss": 0.5004,
+      "step": 1302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99086450382905e-05,
+      "loss": 0.4504,
+      "step": 1303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.990841750157104e-05,
+      "loss": 0.4264,
+      "step": 1304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.990818968236237e-05,
+      "loss": 0.4514,
+      "step": 1305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9907961580667097e-05,
+      "loss": 0.5419,
+      "step": 1306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.990773319648779e-05,
+      "loss": 0.3654,
+      "step": 1307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9907504529827044e-05,
+      "loss": 0.4897,
+      "step": 1308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.990727558068745e-05,
+      "loss": 0.4332,
+      "step": 1309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.990704634907161e-05,
+      "loss": 0.4309,
+      "step": 1310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.990681683498211e-05,
+      "loss": 0.5992,
+      "step": 1311
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.990658703842157e-05,
+      "loss": 0.5554,
+      "step": 1312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9906356959392586e-05,
+      "loss": 0.4939,
+      "step": 1313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.990612659789777e-05,
+      "loss": 0.5205,
+      "step": 1314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.990589595393974e-05,
+      "loss": 0.4859,
+      "step": 1315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9905665027521107e-05,
+      "loss": 0.515,
+      "step": 1316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9905433818644484e-05,
+      "loss": 0.447,
+      "step": 1317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99052023273125e-05,
+      "loss": 0.512,
+      "step": 1318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.990497055352778e-05,
+      "loss": 0.4976,
+      "step": 1319
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.990473849729295e-05,
+      "loss": 0.5009,
+      "step": 1320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9904506158610646e-05,
+      "loss": 0.6271,
+      "step": 1321
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9904273537483505e-05,
+      "loss": 0.5607,
+      "step": 1322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.990404063391415e-05,
+      "loss": 0.4934,
+      "step": 1323
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9903807447905236e-05,
+      "loss": 0.4786,
+      "step": 1324
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9903573979459414e-05,
+      "loss": 0.6906,
+      "step": 1325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.990334022857932e-05,
+      "loss": 0.5004,
+      "step": 1326
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.990310619526759e-05,
+      "loss": 0.4536,
+      "step": 1327
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9902871879526913e-05,
+      "loss": 0.5383,
+      "step": 1328
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9902637281359924e-05,
+      "loss": 0.5234,
+      "step": 1329
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9902402400769286e-05,
+      "loss": 0.5929,
+      "step": 1330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.990216723775768e-05,
+      "loss": 0.4754,
+      "step": 1331
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.990193179232774e-05,
+      "loss": 0.467,
+      "step": 1332
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9901696064482165e-05,
+      "loss": 0.441,
+      "step": 1333
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9901460054223614e-05,
+      "loss": 0.4986,
+      "step": 1334
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9901223761554765e-05,
+      "loss": 0.5093,
+      "step": 1335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.990098718647831e-05,
+      "loss": 0.4958,
+      "step": 1336
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.990075032899691e-05,
+      "loss": 0.5227,
+      "step": 1337
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9900513189113275e-05,
+      "loss": 0.3513,
+      "step": 1338
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.990027576683007e-05,
+      "loss": 0.4795,
+      "step": 1339
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.990003806215001e-05,
+      "loss": 0.3797,
+      "step": 1340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.989980007507579e-05,
+      "loss": 0.5055,
+      "step": 1341
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.989956180561009e-05,
+      "loss": 0.446,
+      "step": 1342
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.989932325375562e-05,
+      "loss": 0.3997,
+      "step": 1343
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.98990844195151e-05,
+      "loss": 0.4522,
+      "step": 1344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.989884530289121e-05,
+      "loss": 0.5041,
+      "step": 1345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.989860590388669e-05,
+      "loss": 0.4781,
+      "step": 1346
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.989836622250424e-05,
+      "loss": 0.4328,
+      "step": 1347
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.989812625874659e-05,
+      "loss": 0.3668,
+      "step": 1348
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9897886012616444e-05,
+      "loss": 0.4793,
+      "step": 1349
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.989764548411654e-05,
+      "loss": 0.6464,
+      "step": 1350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.98974046732496e-05,
+      "loss": 0.4125,
+      "step": 1351
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9897163580018366e-05,
+      "loss": 0.4948,
+      "step": 1352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9896922204425554e-05,
+      "loss": 0.6709,
+      "step": 1353
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9896680546473916e-05,
+      "loss": 0.537,
+      "step": 1354
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.989643860616618e-05,
+      "loss": 0.5108,
+      "step": 1355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.989619638350511e-05,
+      "loss": 0.4417,
+      "step": 1356
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9895953878493434e-05,
+      "loss": 0.5304,
+      "step": 1357
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9895711091133915e-05,
+      "loss": 0.7321,
+      "step": 1358
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9895468021429286e-05,
+      "loss": 0.5395,
+      "step": 1359
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9895224669382324e-05,
+      "loss": 0.4431,
+      "step": 1360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9894981034995794e-05,
+      "loss": 0.8552,
+      "step": 1361
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.989473711827243e-05,
+      "loss": 0.4838,
+      "step": 1362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.989449291921503e-05,
+      "loss": 1.1119,
+      "step": 1363
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.989424843782634e-05,
+      "loss": 0.8745,
+      "step": 1364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.989400367410915e-05,
+      "loss": 0.549,
+      "step": 1365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.989375862806622e-05,
+      "loss": 0.4804,
+      "step": 1366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.989351329970034e-05,
+      "loss": 0.6692,
+      "step": 1367
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.98932676890143e-05,
+      "loss": 0.5109,
+      "step": 1368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.989302179601087e-05,
+      "loss": 0.5942,
+      "step": 1369
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9892775620692844e-05,
+      "loss": 0.7749,
+      "step": 1370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.989252916306301e-05,
+      "loss": 0.4923,
+      "step": 1371
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.989228242312417e-05,
+      "loss": 0.5463,
+      "step": 1372
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.989203540087911e-05,
+      "loss": 0.5924,
+      "step": 1373
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9891788096330654e-05,
+      "loss": 0.718,
+      "step": 1374
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.989154050948159e-05,
+      "loss": 0.5975,
+      "step": 1375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.989129264033472e-05,
+      "loss": 0.5584,
+      "step": 1376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.989104448889288e-05,
+      "loss": 0.8942,
+      "step": 1377
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.989079605515886e-05,
+      "loss": 0.7925,
+      "step": 1378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9890547339135486e-05,
+      "loss": 0.5028,
+      "step": 1379
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.989029834082558e-05,
+      "loss": 0.6466,
+      "step": 1380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.989004906023196e-05,
+      "loss": 0.4568,
+      "step": 1381
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988979949735747e-05,
+      "loss": 0.4817,
+      "step": 1382
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988954965220491e-05,
+      "loss": 0.4067,
+      "step": 1383
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988929952477715e-05,
+      "loss": 0.5958,
+      "step": 1384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9889049115077005e-05,
+      "loss": 0.4873,
+      "step": 1385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988879842310731e-05,
+      "loss": 0.4989,
+      "step": 1386
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9888547448870925e-05,
+      "loss": 0.4938,
+      "step": 1387
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9888296192370686e-05,
+      "loss": 0.5173,
+      "step": 1388
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9888044653609444e-05,
+      "loss": 0.5326,
+      "step": 1389
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988779283259005e-05,
+      "loss": 0.57,
+      "step": 1390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988754072931536e-05,
+      "loss": 0.9365,
+      "step": 1391
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9887288343788244e-05,
+      "loss": 1.6389,
+      "step": 1392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988703567601155e-05,
+      "loss": 0.5587,
+      "step": 1393
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9886782725988153e-05,
+      "loss": 0.8757,
+      "step": 1394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988652949372091e-05,
+      "loss": 0.7831,
+      "step": 1395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988627597921271e-05,
+      "loss": 0.5523,
+      "step": 1396
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988602218246641e-05,
+      "loss": 0.485,
+      "step": 1397
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988576810348491e-05,
+      "loss": 0.5547,
+      "step": 1398
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988551374227107e-05,
+      "loss": 0.5042,
+      "step": 1399
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988525909882779e-05,
+      "loss": 0.7997,
+      "step": 1400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9885004173157934e-05,
+      "loss": 0.7368,
+      "step": 1401
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988474896526443e-05,
+      "loss": 0.623,
+      "step": 1402
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988449347515014e-05,
+      "loss": 0.3859,
+      "step": 1403
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988423770281798e-05,
+      "loss": 0.4781,
+      "step": 1404
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988398164827085e-05,
+      "loss": 0.4184,
+      "step": 1405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988372531151164e-05,
+      "loss": 0.418,
+      "step": 1406
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9883468692543276e-05,
+      "loss": 0.5294,
+      "step": 1407
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988321179136865e-05,
+      "loss": 0.5204,
+      "step": 1408
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988295460799069e-05,
+      "loss": 0.8115,
+      "step": 1409
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988269714241231e-05,
+      "loss": 0.4281,
+      "step": 1410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988243939463642e-05,
+      "loss": 0.4654,
+      "step": 1411
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9882181364665956e-05,
+      "loss": 0.5288,
+      "step": 1412
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988192305250383e-05,
+      "loss": 1.0604,
+      "step": 1413
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988166445815299e-05,
+      "loss": 0.7934,
+      "step": 1414
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988140558161635e-05,
+      "loss": 0.7151,
+      "step": 1415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9881146422896864e-05,
+      "loss": 0.9143,
+      "step": 1416
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9880886981997456e-05,
+      "loss": 0.8296,
+      "step": 1417
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988062725892108e-05,
+      "loss": 0.4611,
+      "step": 1418
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988036725367066e-05,
+      "loss": 0.5172,
+      "step": 1419
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9880106966249176e-05,
+      "loss": 0.3644,
+      "step": 1420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9879846396659555e-05,
+      "loss": 0.5205,
+      "step": 1421
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9879585544904774e-05,
+      "loss": 0.4483,
+      "step": 1422
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9879324410987766e-05,
+      "loss": 0.6036,
+      "step": 1423
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987906299491151e-05,
+      "loss": 0.4928,
+      "step": 1424
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9878801296678965e-05,
+      "loss": 0.51,
+      "step": 1425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987853931629311e-05,
+      "loss": 0.5615,
+      "step": 1426
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9878277053756895e-05,
+      "loss": 0.4449,
+      "step": 1427
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987801450907331e-05,
+      "loss": 0.5232,
+      "step": 1428
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987775168224533e-05,
+      "loss": 0.5281,
+      "step": 1429
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987748857327593e-05,
+      "loss": 0.4304,
+      "step": 1430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9877225182168095e-05,
+      "loss": 0.4497,
+      "step": 1431
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987696150892482e-05,
+      "loss": 0.5591,
+      "step": 1432
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987669755354909e-05,
+      "loss": 0.4703,
+      "step": 1433
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.98764333160439e-05,
+      "loss": 0.4831,
+      "step": 1434
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987616879641225e-05,
+      "loss": 0.4887,
+      "step": 1435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9875903994657125e-05,
+      "loss": 0.3688,
+      "step": 1436
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9875638910781546e-05,
+      "loss": 0.5442,
+      "step": 1437
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987537354478851e-05,
+      "loss": 0.6071,
+      "step": 1438
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987510789668103e-05,
+      "loss": 0.4485,
+      "step": 1439
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987484196646212e-05,
+      "loss": 0.4543,
+      "step": 1440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987457575413479e-05,
+      "loss": 0.5563,
+      "step": 1441
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987430925970206e-05,
+      "loss": 0.4596,
+      "step": 1442
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987404248316696e-05,
+      "loss": 0.5981,
+      "step": 1443
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987377542453251e-05,
+      "loss": 0.488,
+      "step": 1444
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987350808380173e-05,
+      "loss": 0.5869,
+      "step": 1445
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987324046097767e-05,
+      "loss": 0.3748,
+      "step": 1446
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987297255606336e-05,
+      "loss": 0.3965,
+      "step": 1447
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987270436906183e-05,
+      "loss": 0.5811,
+      "step": 1448
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987243589997612e-05,
+      "loss": 0.4977,
+      "step": 1449
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987216714880929e-05,
+      "loss": 0.4727,
+      "step": 1450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987189811556438e-05,
+      "loss": 0.3597,
+      "step": 1451
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987162880024443e-05,
+      "loss": 0.4078,
+      "step": 1452
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9871359202852505e-05,
+      "loss": 0.4953,
+      "step": 1453
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987108932339167e-05,
+      "loss": 0.5821,
+      "step": 1454
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9870819161864976e-05,
+      "loss": 0.419,
+      "step": 1455
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987054871827549e-05,
+      "loss": 0.4378,
+      "step": 1456
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987027799262627e-05,
+      "loss": 0.4588,
+      "step": 1457
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.98700069849204e-05,
+      "loss": 0.5718,
+      "step": 1458
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9869735695160946e-05,
+      "loss": 0.4299,
+      "step": 1459
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9869464123350985e-05,
+      "loss": 0.6344,
+      "step": 1460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.98691922694936e-05,
+      "loss": 0.3483,
+      "step": 1461
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.986892013359187e-05,
+      "loss": 0.5936,
+      "step": 1462
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.986864771564889e-05,
+      "loss": 0.4717,
+      "step": 1463
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9868375015667735e-05,
+      "loss": 0.5495,
+      "step": 1464
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.986810203365151e-05,
+      "loss": 0.3922,
+      "step": 1465
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.986782876960331e-05,
+      "loss": 0.4754,
+      "step": 1466
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.986755522352622e-05,
+      "loss": 0.3417,
+      "step": 1467
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.986728139542336e-05,
+      "loss": 0.3974,
+      "step": 1468
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.986700728529783e-05,
+      "loss": 0.603,
+      "step": 1469
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9866732893152735e-05,
+      "loss": 0.4104,
+      "step": 1470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.98664582189912e-05,
+      "loss": 0.4542,
+      "step": 1471
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.986618326281631e-05,
+      "loss": 0.4422,
+      "step": 1472
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.986590802463121e-05,
+      "loss": 0.4559,
+      "step": 1473
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.986563250443902e-05,
+      "loss": 0.519,
+      "step": 1474
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.986535670224286e-05,
+      "loss": 0.4281,
+      "step": 1475
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9865080618045854e-05,
+      "loss": 0.4309,
+      "step": 1476
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9864804251851134e-05,
+      "loss": 0.6023,
+      "step": 1477
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.986452760366184e-05,
+      "loss": 0.3576,
+      "step": 1478
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.98642506734811e-05,
+      "loss": 0.4123,
+      "step": 1479
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.986397346131206e-05,
+      "loss": 0.3861,
+      "step": 1480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9863695967157874e-05,
+      "loss": 0.3498,
+      "step": 1481
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9863418191021675e-05,
+      "loss": 0.7751,
+      "step": 1482
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9863140132906615e-05,
+      "loss": 0.4906,
+      "step": 1483
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.986286179281585e-05,
+      "loss": 0.4467,
+      "step": 1484
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.986258317075254e-05,
+      "loss": 0.4595,
+      "step": 1485
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9862304266719836e-05,
+      "loss": 0.4545,
+      "step": 1486
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.986202508072091e-05,
+      "loss": 0.3859,
+      "step": 1487
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.986174561275893e-05,
+      "loss": 0.4755,
+      "step": 1488
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.986146586283705e-05,
+      "loss": 0.4318,
+      "step": 1489
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9861185830958466e-05,
+      "loss": 0.4383,
+      "step": 1490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9860905517126336e-05,
+      "loss": 0.3834,
+      "step": 1491
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9860624921343846e-05,
+      "loss": 0.4114,
+      "step": 1492
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.986034404361417e-05,
+      "loss": 0.3468,
+      "step": 1493
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.986006288394051e-05,
+      "loss": 0.4396,
+      "step": 1494
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9859781442326034e-05,
+      "loss": 0.3947,
+      "step": 1495
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.985949971877395e-05,
+      "loss": 0.5733,
+      "step": 1496
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.985921771328745e-05,
+      "loss": 0.5674,
+      "step": 1497
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.985893542586972e-05,
+      "loss": 0.6103,
+      "step": 1498
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9858652856523985e-05,
+      "loss": 0.437,
+      "step": 1499
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.985837000525343e-05,
+      "loss": 0.3957,
+      "step": 1500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.985808687206127e-05,
+      "loss": 0.4393,
+      "step": 1501
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.985780345695071e-05,
+      "loss": 0.5017,
+      "step": 1502
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9857519759924974e-05,
+      "loss": 0.4982,
+      "step": 1503
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.985723578098728e-05,
+      "loss": 0.4252,
+      "step": 1504
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.985695152014083e-05,
+      "loss": 0.4999,
+      "step": 1505
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9856666977388865e-05,
+      "loss": 0.4434,
+      "step": 1506
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.985638215273461e-05,
+      "loss": 0.4868,
+      "step": 1507
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.985609704618129e-05,
+      "loss": 0.5155,
+      "step": 1508
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.985581165773216e-05,
+      "loss": 0.6882,
+      "step": 1509
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.985552598739042e-05,
+      "loss": 0.3738,
+      "step": 1510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.985524003515932e-05,
+      "loss": 0.4722,
+      "step": 1511
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.985495380104213e-05,
+      "loss": 0.501,
+      "step": 1512
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.985466728504207e-05,
+      "loss": 0.5022,
+      "step": 1513
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.98543804871624e-05,
+      "loss": 0.3599,
+      "step": 1514
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.985409340740636e-05,
+      "loss": 0.5191,
+      "step": 1515
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.985380604577722e-05,
+      "loss": 0.4497,
+      "step": 1516
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.985351840227824e-05,
+      "loss": 0.481,
+      "step": 1517
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.985323047691267e-05,
+      "loss": 0.3634,
+      "step": 1518
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.985294226968379e-05,
+      "loss": 0.4295,
+      "step": 1519
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.985265378059485e-05,
+      "loss": 0.413,
+      "step": 1520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.985236500964914e-05,
+      "loss": 0.3875,
+      "step": 1521
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.985207595684992e-05,
+      "loss": 0.3282,
+      "step": 1522
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9851786622200484e-05,
+      "loss": 0.4027,
+      "step": 1523
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9851497005704097e-05,
+      "loss": 0.3616,
+      "step": 1524
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.985120710736405e-05,
+      "loss": 0.4686,
+      "step": 1525
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.985091692718364e-05,
+      "loss": 0.4145,
+      "step": 1526
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9850626465166145e-05,
+      "loss": 0.6516,
+      "step": 1527
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.985033572131487e-05,
+      "loss": 0.4088,
+      "step": 1528
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9850044695633105e-05,
+      "loss": 0.4226,
+      "step": 1529
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.984975338812414e-05,
+      "loss": 0.4567,
+      "step": 1530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.98494617987913e-05,
+      "loss": 0.4334,
+      "step": 1531
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.984916992763788e-05,
+      "loss": 0.5841,
+      "step": 1532
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.98488777746672e-05,
+      "loss": 0.4446,
+      "step": 1533
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.984858533988256e-05,
+      "loss": 0.4903,
+      "step": 1534
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.984829262328728e-05,
+      "loss": 0.4427,
+      "step": 1535
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.98479996248847e-05,
+      "loss": 0.3805,
+      "step": 1536
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9847706344678106e-05,
+      "loss": 0.4537,
+      "step": 1537
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.984741278267086e-05,
+      "loss": 0.4172,
+      "step": 1538
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9847118938866255e-05,
+      "loss": 0.4076,
+      "step": 1539
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.984682481326766e-05,
+      "loss": 0.4433,
+      "step": 1540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.984653040587839e-05,
+      "loss": 0.3581,
+      "step": 1541
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.984623571670178e-05,
+      "loss": 0.3236,
+      "step": 1542
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.984594074574119e-05,
+      "loss": 0.4444,
+      "step": 1543
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.984564549299995e-05,
+      "loss": 0.4471,
+      "step": 1544
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.984534995848141e-05,
+      "loss": 0.3658,
+      "step": 1545
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9845054142188936e-05,
+      "loss": 0.4334,
+      "step": 1546
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.984475804412587e-05,
+      "loss": 0.4344,
+      "step": 1547
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.984446166429557e-05,
+      "loss": 0.3635,
+      "step": 1548
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9844165002701395e-05,
+      "loss": 0.4519,
+      "step": 1549
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.984386805934672e-05,
+      "loss": 0.4095,
+      "step": 1550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.984357083423491e-05,
+      "loss": 0.4776,
+      "step": 1551
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9843273327369334e-05,
+      "loss": 0.724,
+      "step": 1552
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9842975538753357e-05,
+      "loss": 0.3415,
+      "step": 1553
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.984267746839036e-05,
+      "loss": 0.6649,
+      "step": 1554
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9842379116283744e-05,
+      "loss": 0.4228,
+      "step": 1555
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.984208048243687e-05,
+      "loss": 0.4171,
+      "step": 1556
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.984178156685313e-05,
+      "loss": 0.5316,
+      "step": 1557
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.984148236953591e-05,
+      "loss": 0.3471,
+      "step": 1558
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9841182890488615e-05,
+      "loss": 0.3927,
+      "step": 1559
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.984088312971463e-05,
+      "loss": 0.5258,
+      "step": 1560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.984058308721736e-05,
+      "loss": 0.5162,
+      "step": 1561
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.984028276300021e-05,
+      "loss": 0.5791,
+      "step": 1562
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.983998215706659e-05,
+      "loss": 0.8016,
+      "step": 1563
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9839681269419894e-05,
+      "loss": 0.4312,
+      "step": 1564
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.983938010006354e-05,
+      "loss": 0.4251,
+      "step": 1565
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.983907864900095e-05,
+      "loss": 0.5379,
+      "step": 1566
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.983877691623554e-05,
+      "loss": 0.4888,
+      "step": 1567
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.983847490177073e-05,
+      "loss": 0.5421,
+      "step": 1568
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9838172605609946e-05,
+      "loss": 0.4214,
+      "step": 1569
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.983787002775662e-05,
+      "loss": 0.4662,
+      "step": 1570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9837567168214186e-05,
+      "loss": 0.4719,
+      "step": 1571
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.983726402698606e-05,
+      "loss": 0.4297,
+      "step": 1572
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9836960604075697e-05,
+      "loss": 0.4548,
+      "step": 1573
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9836656899486536e-05,
+      "loss": 0.4733,
+      "step": 1574
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9836352913222024e-05,
+      "loss": 0.4194,
+      "step": 1575
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9836048645285596e-05,
+      "loss": 0.49,
+      "step": 1576
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9835744095680715e-05,
+      "loss": 0.4562,
+      "step": 1577
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9835439264410835e-05,
+      "loss": 0.3933,
+      "step": 1578
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.98351341514794e-05,
+      "loss": 0.4119,
+      "step": 1579
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.983482875688988e-05,
+      "loss": 0.4109,
+      "step": 1580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.983452308064575e-05,
+      "loss": 0.613,
+      "step": 1581
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.983421712275046e-05,
+      "loss": 0.4638,
+      "step": 1582
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.983391088320748e-05,
+      "loss": 0.3608,
+      "step": 1583
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9833604362020283e-05,
+      "loss": 0.4817,
+      "step": 1584
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.983329755919236e-05,
+      "loss": 0.469,
+      "step": 1585
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.983299047472717e-05,
+      "loss": 0.4486,
+      "step": 1586
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9832683108628216e-05,
+      "loss": 0.549,
+      "step": 1587
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9832375460898964e-05,
+      "loss": 0.4985,
+      "step": 1588
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9832067531542925e-05,
+      "loss": 0.4218,
+      "step": 1589
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.983175932056357e-05,
+      "loss": 0.507,
+      "step": 1590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.983145082796441e-05,
+      "loss": 0.5638,
+      "step": 1591
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.983114205374893e-05,
+      "loss": 0.5026,
+      "step": 1592
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.983083299792064e-05,
+      "loss": 0.4008,
+      "step": 1593
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.983052366048305e-05,
+      "loss": 0.4396,
+      "step": 1594
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.983021404143966e-05,
+      "loss": 0.3454,
+      "step": 1595
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.982990414079398e-05,
+      "loss": 0.4204,
+      "step": 1596
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.982959395854953e-05,
+      "loss": 0.5109,
+      "step": 1597
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9829283494709824e-05,
+      "loss": 0.3948,
+      "step": 1598
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.982897274927839e-05,
+      "loss": 0.4392,
+      "step": 1599
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.982866172225875e-05,
+      "loss": 0.3867,
+      "step": 1600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9828350413654424e-05,
+      "loss": 0.3923,
+      "step": 1601
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.982803882346895e-05,
+      "loss": 0.4158,
+      "step": 1602
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.982772695170586e-05,
+      "loss": 0.3835,
+      "step": 1603
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.982741479836869e-05,
+      "loss": 0.6963,
+      "step": 1604
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.982710236346098e-05,
+      "loss": 0.5531,
+      "step": 1605
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.982678964698626e-05,
+      "loss": 0.3971,
+      "step": 1606
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.982647664894811e-05,
+      "loss": 0.4154,
+      "step": 1607
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.982616336935005e-05,
+      "loss": 0.408,
+      "step": 1608
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.982584980819565e-05,
+      "loss": 0.3993,
+      "step": 1609
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.982553596548845e-05,
+      "loss": 0.482,
+      "step": 1610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9825221841232014e-05,
+      "loss": 0.3883,
+      "step": 1611
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.982490743542992e-05,
+      "loss": 0.5074,
+      "step": 1612
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9824592748085725e-05,
+      "loss": 0.4577,
+      "step": 1613
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.982427777920299e-05,
+      "loss": 0.4983,
+      "step": 1614
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.982396252878529e-05,
+      "loss": 0.4059,
+      "step": 1615
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.982364699683621e-05,
+      "loss": 0.2965,
+      "step": 1616
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9823331183359315e-05,
+      "loss": 0.3866,
+      "step": 1617
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9823015088358196e-05,
+      "loss": 0.4504,
+      "step": 1618
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9822698711836434e-05,
+      "loss": 0.4343,
+      "step": 1619
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.982238205379762e-05,
+      "loss": 0.4673,
+      "step": 1620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.982206511424534e-05,
+      "loss": 0.3491,
+      "step": 1621
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.98217478931832e-05,
+      "loss": 0.4092,
+      "step": 1622
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.982143039061478e-05,
+      "loss": 0.3955,
+      "step": 1623
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9821112606543694e-05,
+      "loss": 0.4634,
+      "step": 1624
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.982079454097354e-05,
+      "loss": 0.4651,
+      "step": 1625
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.982047619390793e-05,
+      "loss": 0.3641,
+      "step": 1626
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.982015756535047e-05,
+      "loss": 0.525,
+      "step": 1627
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.981983865530478e-05,
+      "loss": 0.2956,
+      "step": 1628
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9819519463774464e-05,
+      "loss": 0.467,
+      "step": 1629
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.981919999076315e-05,
+      "loss": 0.4708,
+      "step": 1630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.981888023627447e-05,
+      "loss": 0.238,
+      "step": 1631
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9818560200312045e-05,
+      "loss": 0.3458,
+      "step": 1632
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9818239882879494e-05,
+      "loss": 0.4263,
+      "step": 1633
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9817919283980455e-05,
+      "loss": 0.4382,
+      "step": 1634
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.981759840361857e-05,
+      "loss": 0.4194,
+      "step": 1635
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.981727724179747e-05,
+      "loss": 0.5317,
+      "step": 1636
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.981695579852081e-05,
+      "loss": 0.4356,
+      "step": 1637
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9816634073792216e-05,
+      "loss": 0.3925,
+      "step": 1638
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.981631206761536e-05,
+      "loss": 0.4757,
+      "step": 1639
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.981598977999386e-05,
+      "loss": 0.5004,
+      "step": 1640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.981566721093142e-05,
+      "loss": 0.4617,
+      "step": 1641
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.981534436043166e-05,
+      "loss": 0.395,
+      "step": 1642
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.981502122849825e-05,
+      "loss": 0.4801,
+      "step": 1643
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.981469781513486e-05,
+      "loss": 0.5404,
+      "step": 1644
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9814374120345144e-05,
+      "loss": 0.3718,
+      "step": 1645
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.98140501441328e-05,
+      "loss": 0.8161,
+      "step": 1646
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.981372588650148e-05,
+      "loss": 0.5007,
+      "step": 1647
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.981340134745487e-05,
+      "loss": 0.3762,
+      "step": 1648
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9813076526996646e-05,
+      "loss": 0.4682,
+      "step": 1649
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.981275142513049e-05,
+      "loss": 0.377,
+      "step": 1650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.98124260418601e-05,
+      "loss": 0.4539,
+      "step": 1651
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9812100377189156e-05,
+      "loss": 0.4523,
+      "step": 1652
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.981177443112136e-05,
+      "loss": 0.4278,
+      "step": 1653
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.98114482036604e-05,
+      "loss": 0.4444,
+      "step": 1654
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.981112169480998e-05,
+      "loss": 0.4127,
+      "step": 1655
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.98107949045738e-05,
+      "loss": 0.3829,
+      "step": 1656
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.981046783295557e-05,
+      "loss": 0.3925,
+      "step": 1657
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9810140479959e-05,
+      "loss": 0.4359,
+      "step": 1658
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.98098128455878e-05,
+      "loss": 0.4024,
+      "step": 1659
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9809484929845686e-05,
+      "loss": 0.3839,
+      "step": 1660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980915673273637e-05,
+      "loss": 0.423,
+      "step": 1661
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980882825426358e-05,
+      "loss": 0.5112,
+      "step": 1662
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980849949443106e-05,
+      "loss": 0.3869,
+      "step": 1663
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980817045324251e-05,
+      "loss": 0.3948,
+      "step": 1664
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980784113070167e-05,
+      "loss": 0.4176,
+      "step": 1665
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980751152681228e-05,
+      "loss": 0.4821,
+      "step": 1666
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980718164157808e-05,
+      "loss": 0.3199,
+      "step": 1667
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.98068514750028e-05,
+      "loss": 0.3843,
+      "step": 1668
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980652102709019e-05,
+      "loss": 0.4408,
+      "step": 1669
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980619029784401e-05,
+      "loss": 0.411,
+      "step": 1670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980585928726799e-05,
+      "loss": 0.4563,
+      "step": 1671
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9805527995365896e-05,
+      "loss": 0.3939,
+      "step": 1672
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9805196422141486e-05,
+      "loss": 0.4388,
+      "step": 1673
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9804864567598516e-05,
+      "loss": 0.5022,
+      "step": 1674
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9804532431740745e-05,
+      "loss": 0.4715,
+      "step": 1675
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9804200014571955e-05,
+      "loss": 0.462,
+      "step": 1676
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980386731609591e-05,
+      "loss": 0.3784,
+      "step": 1677
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980353433631637e-05,
+      "loss": 0.3052,
+      "step": 1678
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980320107523712e-05,
+      "loss": 0.4557,
+      "step": 1679
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980286753286195e-05,
+      "loss": 0.3425,
+      "step": 1680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9802533709194625e-05,
+      "loss": 0.3406,
+      "step": 1681
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980219960423895e-05,
+      "loss": 0.3844,
+      "step": 1682
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9801865217998694e-05,
+      "loss": 0.4601,
+      "step": 1683
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980153055047766e-05,
+      "loss": 0.4541,
+      "step": 1684
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980119560167965e-05,
+      "loss": 0.4741,
+      "step": 1685
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980086037160845e-05,
+      "loss": 0.5092,
+      "step": 1686
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980052486026787e-05,
+      "loss": 0.3317,
+      "step": 1687
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9800189067661705e-05,
+      "loss": 0.3661,
+      "step": 1688
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.979985299379378e-05,
+      "loss": 0.3621,
+      "step": 1689
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9799516638667884e-05,
+      "loss": 0.4123,
+      "step": 1690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.979918000228786e-05,
+      "loss": 0.2914,
+      "step": 1691
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.97988430846575e-05,
+      "loss": 0.3786,
+      "step": 1692
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.979850588578065e-05,
+      "loss": 0.3585,
+      "step": 1693
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.979816840566109e-05,
+      "loss": 0.4777,
+      "step": 1694
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.97978306443027e-05,
+      "loss": 0.2357,
+      "step": 1695
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9797492601709285e-05,
+      "loss": 0.4411,
+      "step": 1696
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.979715427788468e-05,
+      "loss": 0.4962,
+      "step": 1697
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.979681567283272e-05,
+      "loss": 0.2725,
+      "step": 1698
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.979647678655726e-05,
+      "loss": 0.3152,
+      "step": 1699
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.979613761906212e-05,
+      "loss": 0.3196,
+      "step": 1700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.979579817035116e-05,
+      "loss": 0.3359,
+      "step": 1701
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9795458440428225e-05,
+      "loss": 0.4443,
+      "step": 1702
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9795118429297185e-05,
+      "loss": 0.5231,
+      "step": 1703
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9794778136961866e-05,
+      "loss": 0.58,
+      "step": 1704
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.979443756342616e-05,
+      "loss": 0.7294,
+      "step": 1705
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9794096708693894e-05,
+      "loss": 0.4221,
+      "step": 1706
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9793755572768966e-05,
+      "loss": 0.4171,
+      "step": 1707
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.979341415565523e-05,
+      "loss": 0.4804,
+      "step": 1708
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.979307245735655e-05,
+      "loss": 0.5104,
+      "step": 1709
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9792730477876816e-05,
+      "loss": 0.4927,
+      "step": 1710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.979238821721991e-05,
+      "loss": 0.5054,
+      "step": 1711
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.979204567538969e-05,
+      "loss": 0.3932,
+      "step": 1712
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.979170285239006e-05,
+      "loss": 0.4152,
+      "step": 1713
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.979135974822491e-05,
+      "loss": 0.3172,
+      "step": 1714
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.979101636289812e-05,
+      "loss": 0.4826,
+      "step": 1715
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.979067269641359e-05,
+      "loss": 0.4654,
+      "step": 1716
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9790328748775214e-05,
+      "loss": 0.4268,
+      "step": 1717
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9789984519986905e-05,
+      "loss": 0.7405,
+      "step": 1718
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.978964001005255e-05,
+      "loss": 0.3812,
+      "step": 1719
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.978929521897607e-05,
+      "loss": 0.345,
+      "step": 1720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.978895014676136e-05,
+      "loss": 0.3702,
+      "step": 1721
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9788604793412344e-05,
+      "loss": 0.3971,
+      "step": 1722
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.978825915893294e-05,
+      "loss": 0.3506,
+      "step": 1723
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.978791324332707e-05,
+      "loss": 0.2984,
+      "step": 1724
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.978756704659864e-05,
+      "loss": 0.4043,
+      "step": 1725
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9787220568751594e-05,
+      "loss": 0.3489,
+      "step": 1726
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.978687380978986e-05,
+      "loss": 0.3344,
+      "step": 1727
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.978652676971736e-05,
+      "loss": 0.315,
+      "step": 1728
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.978617944853804e-05,
+      "loss": 0.2808,
+      "step": 1729
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9785831846255823e-05,
+      "loss": 0.3607,
+      "step": 1730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.978548396287468e-05,
+      "loss": 0.5055,
+      "step": 1731
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.978513579839853e-05,
+      "loss": 0.4461,
+      "step": 1732
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.978478735283134e-05,
+      "loss": 0.2809,
+      "step": 1733
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.978443862617704e-05,
+      "loss": 0.3156,
+      "step": 1734
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.978408961843961e-05,
+      "loss": 0.6265,
+      "step": 1735
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9783740329623e-05,
+      "loss": 0.3929,
+      "step": 1736
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.978339075973115e-05,
+      "loss": 0.6448,
+      "step": 1737
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.978304090876805e-05,
+      "loss": 0.4029,
+      "step": 1738
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.978269077673767e-05,
+      "loss": 0.4841,
+      "step": 1739
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.978234036364396e-05,
+      "loss": 0.4755,
+      "step": 1740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.97819896694909e-05,
+      "loss": 1.4906,
+      "step": 1741
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.978163869428248e-05,
+      "loss": 1.7674,
+      "step": 1742
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9781287438022673e-05,
+      "loss": 1.0987,
+      "step": 1743
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.978093590071546e-05,
+      "loss": 2.0869,
+      "step": 1744
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9780584082364825e-05,
+      "loss": 1.7215,
+      "step": 1745
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9780231982974776e-05,
+      "loss": 0.8221,
+      "step": 1746
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9779879602549284e-05,
+      "loss": 1.0816,
+      "step": 1747
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.977952694109235e-05,
+      "loss": 1.7231,
+      "step": 1748
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.977917399860799e-05,
+      "loss": 1.3701,
+      "step": 1749
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9778820775100185e-05,
+      "loss": 2.2941,
+      "step": 1750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9778467270572946e-05,
+      "loss": 1.0376,
+      "step": 1751
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9778113485030296e-05,
+      "loss": 2.0672,
+      "step": 1752
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.977775941847624e-05,
+      "loss": 1.399,
+      "step": 1753
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.977740507091479e-05,
+      "loss": 1.7084,
+      "step": 1754
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.977705044234996e-05,
+      "loss": 2.3192,
+      "step": 1755
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9776695532785784e-05,
+      "loss": 1.7677,
+      "step": 1756
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.977634034222628e-05,
+      "loss": 1.721,
+      "step": 1757
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9775984870675475e-05,
+      "loss": 0.4435,
+      "step": 1758
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9775629118137404e-05,
+      "loss": 2.32,
+      "step": 1759
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.97752730846161e-05,
+      "loss": 0.4293,
+      "step": 1760
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.977491677011561e-05,
+      "loss": 1.3866,
+      "step": 1761
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.977456017463996e-05,
+      "loss": 0.398,
+      "step": 1762
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.97742032981932e-05,
+      "loss": 1.0291,
+      "step": 1763
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9773846140779376e-05,
+      "loss": 0.8506,
+      "step": 1764
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.977348870240254e-05,
+      "loss": 1.3533,
+      "step": 1765
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.977313098306675e-05,
+      "loss": 1.3849,
+      "step": 1766
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9772772982776064e-05,
+      "loss": 0.7473,
+      "step": 1767
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.977241470153453e-05,
+      "loss": 1.7277,
+      "step": 1768
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.977205613934622e-05,
+      "loss": 1.0279,
+      "step": 1769
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.97716972962152e-05,
+      "loss": 1.2154,
+      "step": 1770
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.977133817214554e-05,
+      "loss": 2.1357,
+      "step": 1771
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9770978767141316e-05,
+      "loss": 1.0413,
+      "step": 1772
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.977061908120659e-05,
+      "loss": 2.482,
+      "step": 1773
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.977025911434546e-05,
+      "loss": 1.446,
+      "step": 1774
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.976989886656199e-05,
+      "loss": 0.806,
+      "step": 1775
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.976953833786028e-05,
+      "loss": 1.3376,
+      "step": 1776
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9769177528244415e-05,
+      "loss": 0.6662,
+      "step": 1777
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.976881643771849e-05,
+      "loss": 1.3378,
+      "step": 1778
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.976845506628659e-05,
+      "loss": 2.0527,
+      "step": 1779
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9768093413952824e-05,
+      "loss": 1.3472,
+      "step": 1780
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9767731480721284e-05,
+      "loss": 1.445,
+      "step": 1781
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9767369266596074e-05,
+      "loss": 2.1223,
+      "step": 1782
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9767006771581315e-05,
+      "loss": 0.4601,
+      "step": 1783
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9766643995681114e-05,
+      "loss": 0.5137,
+      "step": 1784
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9766280938899576e-05,
+      "loss": 1.0348,
+      "step": 1785
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.976591760124083e-05,
+      "loss": 1.3887,
+      "step": 1786
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9765553982708985e-05,
+      "loss": 1.0041,
+      "step": 1787
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.976519008330818e-05,
+      "loss": 1.0589,
+      "step": 1788
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.976482590304252e-05,
+      "loss": 1.6761,
+      "step": 1789
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.976446144191616e-05,
+      "loss": 1.0446,
+      "step": 1790
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.976409669993322e-05,
+      "loss": 1.3562,
+      "step": 1791
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.976373167709783e-05,
+      "loss": 2.3881,
+      "step": 1792
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.976336637341414e-05,
+      "loss": 2.136,
+      "step": 1793
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9763000788886295e-05,
+      "loss": 1.3459,
+      "step": 1794
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.976263492351843e-05,
+      "loss": 2.0268,
+      "step": 1795
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.976226877731471e-05,
+      "loss": 1.3035,
+      "step": 1796
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9761902350279275e-05,
+      "loss": 1.4184,
+      "step": 1797
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.976153564241628e-05,
+      "loss": 1.7767,
+      "step": 1798
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.97611686537299e-05,
+      "loss": 0.7229,
+      "step": 1799
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.976080138422428e-05,
+      "loss": 2.2997,
+      "step": 1800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.976043383390359e-05,
+      "loss": 0.6625,
+      "step": 1801
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9760066002771985e-05,
+      "loss": 0.4486,
+      "step": 1802
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.975969789083367e-05,
+      "loss": 0.3876,
+      "step": 1803
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9759329498092794e-05,
+      "loss": 1.0678,
+      "step": 1804
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9758960824553546e-05,
+      "loss": 2.7588,
+      "step": 1805
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.975859187022009e-05,
+      "loss": 1.6837,
+      "step": 1806
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.975822263509664e-05,
+      "loss": 1.6497,
+      "step": 1807
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.975785311918736e-05,
+      "loss": 1.2244,
+      "step": 1808
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.975748332249644e-05,
+      "loss": 1.0043,
+      "step": 1809
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9757113245028086e-05,
+      "loss": 1.1306,
+      "step": 1810
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9756742886786486e-05,
+      "loss": 2.0158,
+      "step": 1811
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.975637224777585e-05,
+      "loss": 1.7204,
+      "step": 1812
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.975600132800038e-05,
+      "loss": 0.7505,
+      "step": 1813
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.975563012746427e-05,
+      "loss": 0.3607,
+      "step": 1814
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.975525864617175e-05,
+      "loss": 0.343,
+      "step": 1815
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.975488688412701e-05,
+      "loss": 0.31,
+      "step": 1816
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.975451484133429e-05,
+      "loss": 0.4082,
+      "step": 1817
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.975414251779779e-05,
+      "loss": 0.4284,
+      "step": 1818
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.975376991352173e-05,
+      "loss": 0.362,
+      "step": 1819
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.975339702851035e-05,
+      "loss": 0.4582,
+      "step": 1820
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9753023862767884e-05,
+      "loss": 0.4656,
+      "step": 1821
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.975265041629855e-05,
+      "loss": 0.3234,
+      "step": 1822
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.975227668910658e-05,
+      "loss": 0.3211,
+      "step": 1823
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.975190268119623e-05,
+      "loss": 0.2979,
+      "step": 1824
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.975152839257173e-05,
+      "loss": 0.3385,
+      "step": 1825
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.975115382323732e-05,
+      "loss": 0.502,
+      "step": 1826
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.975077897319726e-05,
+      "loss": 0.4601,
+      "step": 1827
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9750403842455804e-05,
+      "loss": 0.38,
+      "step": 1828
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.975002843101719e-05,
+      "loss": 0.4174,
+      "step": 1829
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9749652738885676e-05,
+      "loss": 0.3758,
+      "step": 1830
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9749276766065545e-05,
+      "loss": 0.3257,
+      "step": 1831
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9748900512561037e-05,
+      "loss": 0.3884,
+      "step": 1832
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.974852397837644e-05,
+      "loss": 0.3672,
+      "step": 1833
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9748147163516e-05,
+      "loss": 0.3276,
+      "step": 1834
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.974777006798401e-05,
+      "loss": 0.3695,
+      "step": 1835
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.974739269178474e-05,
+      "loss": 0.2754,
+      "step": 1836
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.974701503492247e-05,
+      "loss": 0.3584,
+      "step": 1837
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.974663709740148e-05,
+      "loss": 0.3888,
+      "step": 1838
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.974625887922606e-05,
+      "loss": 0.3279,
+      "step": 1839
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9745880380400504e-05,
+      "loss": 0.4308,
+      "step": 1840
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.97455016009291e-05,
+      "loss": 0.4313,
+      "step": 1841
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9745122540816137e-05,
+      "loss": 0.357,
+      "step": 1842
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9744743200065925e-05,
+      "loss": 0.3427,
+      "step": 1843
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9744363578682755e-05,
+      "loss": 0.4911,
+      "step": 1844
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.974398367667095e-05,
+      "loss": 0.4195,
+      "step": 1845
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9743603494034796e-05,
+      "loss": 0.5349,
+      "step": 1846
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.974322303077862e-05,
+      "loss": 0.2968,
+      "step": 1847
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.974284228690673e-05,
+      "loss": 0.3742,
+      "step": 1848
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.974246126242346e-05,
+      "loss": 0.3208,
+      "step": 1849
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.97420799573331e-05,
+      "loss": 0.4984,
+      "step": 1850
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.974169837164e-05,
+      "loss": 0.3404,
+      "step": 1851
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9741316505348484e-05,
+      "loss": 0.4069,
+      "step": 1852
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9740934358462876e-05,
+      "loss": 0.3007,
+      "step": 1853
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9740551930987514e-05,
+      "loss": 0.4649,
+      "step": 1854
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.974016922292673e-05,
+      "loss": 0.3994,
+      "step": 1855
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.973978623428487e-05,
+      "loss": 0.2856,
+      "step": 1856
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9739402965066276e-05,
+      "loss": 0.3915,
+      "step": 1857
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9739019415275295e-05,
+      "loss": 0.3324,
+      "step": 1858
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.973863558491628e-05,
+      "loss": 0.3643,
+      "step": 1859
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9738251473993576e-05,
+      "loss": 0.3185,
+      "step": 1860
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9737867082511546e-05,
+      "loss": 0.3815,
+      "step": 1861
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9737482410474545e-05,
+      "loss": 0.3655,
+      "step": 1862
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9737097457886936e-05,
+      "loss": 0.435,
+      "step": 1863
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.973671222475309e-05,
+      "loss": 0.3232,
+      "step": 1864
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9736326711077374e-05,
+      "loss": 0.2737,
+      "step": 1865
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9735940916864155e-05,
+      "loss": 0.3769,
+      "step": 1866
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.973555484211782e-05,
+      "loss": 0.3947,
+      "step": 1867
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9735168486842734e-05,
+      "loss": 0.3018,
+      "step": 1868
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.973478185104328e-05,
+      "loss": 0.3758,
+      "step": 1869
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.973439493472385e-05,
+      "loss": 0.3825,
+      "step": 1870
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.973400773788883e-05,
+      "loss": 0.6368,
+      "step": 1871
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9733620260542614e-05,
+      "loss": 0.2649,
+      "step": 1872
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9733232502689586e-05,
+      "loss": 0.3353,
+      "step": 1873
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.973284446433416e-05,
+      "loss": 0.4147,
+      "step": 1874
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.973245614548072e-05,
+      "loss": 0.2996,
+      "step": 1875
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9732067546133675e-05,
+      "loss": 0.3889,
+      "step": 1876
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.973167866629744e-05,
+      "loss": 0.4952,
+      "step": 1877
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.973128950597641e-05,
+      "loss": 0.4744,
+      "step": 1878
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9730900065175027e-05,
+      "loss": 0.4312,
+      "step": 1879
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9730510343897675e-05,
+      "loss": 0.4335,
+      "step": 1880
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.973012034214879e-05,
+      "loss": 0.6588,
+      "step": 1881
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9729730059932787e-05,
+      "loss": 0.3337,
+      "step": 1882
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.972933949725411e-05,
+      "loss": 0.3494,
+      "step": 1883
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.972894865411717e-05,
+      "loss": 0.3089,
+      "step": 1884
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.972855753052641e-05,
+      "loss": 0.4534,
+      "step": 1885
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9728166126486254e-05,
+      "loss": 0.4386,
+      "step": 1886
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9727774442001156e-05,
+      "loss": 0.348,
+      "step": 1887
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9727382477075544e-05,
+      "loss": 0.5509,
+      "step": 1888
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.972699023171388e-05,
+      "loss": 0.3722,
+      "step": 1889
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9726597705920597e-05,
+      "loss": 0.3535,
+      "step": 1890
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.972620489970016e-05,
+      "loss": 0.4308,
+      "step": 1891
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9725811813057007e-05,
+      "loss": 0.5229,
+      "step": 1892
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.972541844599561e-05,
+      "loss": 0.5085,
+      "step": 1893
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.972502479852042e-05,
+      "loss": 0.4069,
+      "step": 1894
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.972463087063591e-05,
+      "loss": 0.4082,
+      "step": 1895
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.972423666234655e-05,
+      "loss": 0.3436,
+      "step": 1896
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.97238421736568e-05,
+      "loss": 0.4126,
+      "step": 1897
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.972344740457114e-05,
+      "loss": 0.3783,
+      "step": 1898
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.972305235509405e-05,
+      "loss": 0.3647,
+      "step": 1899
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.972265702523e-05,
+      "loss": 0.3417,
+      "step": 1900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.972226141498349e-05,
+      "loss": 0.3148,
+      "step": 1901
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.972186552435898e-05,
+      "loss": 0.5203,
+      "step": 1902
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.972146935336099e-05,
+      "loss": 0.3376,
+      "step": 1903
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.972107290199399e-05,
+      "loss": 0.4464,
+      "step": 1904
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.97206761702625e-05,
+      "loss": 0.3561,
+      "step": 1905
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9720279158171e-05,
+      "loss": 0.3964,
+      "step": 1906
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9719881865723996e-05,
+      "loss": 0.3548,
+      "step": 1907
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.971948429292599e-05,
+      "loss": 0.3893,
+      "step": 1908
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.97190864397815e-05,
+      "loss": 0.3625,
+      "step": 1909
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.971868830629504e-05,
+      "loss": 0.3352,
+      "step": 1910
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.971828989247111e-05,
+      "loss": 0.5455,
+      "step": 1911
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.971789119831425e-05,
+      "loss": 0.6604,
+      "step": 1912
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.971749222382896e-05,
+      "loss": 0.4307,
+      "step": 1913
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.971709296901978e-05,
+      "loss": 0.3181,
+      "step": 1914
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.971669343389124e-05,
+      "loss": 0.32,
+      "step": 1915
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.971629361844785e-05,
+      "loss": 0.3843,
+      "step": 1916
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.971589352269417e-05,
+      "loss": 0.3169,
+      "step": 1917
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9715493146634715e-05,
+      "loss": 0.2852,
+      "step": 1918
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9715092490274046e-05,
+      "loss": 0.4507,
+      "step": 1919
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.971469155361669e-05,
+      "loss": 0.3661,
+      "step": 1920
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.971429033666721e-05,
+      "loss": 0.4866,
+      "step": 1921
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.971388883943014e-05,
+      "loss": 0.2552,
+      "step": 1922
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.971348706191005e-05,
+      "loss": 0.4241,
+      "step": 1923
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.971308500411147e-05,
+      "loss": 0.4878,
+      "step": 1924
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.971268266603899e-05,
+      "loss": 0.3364,
+      "step": 1925
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9712280047697164e-05,
+      "loss": 0.4274,
+      "step": 1926
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.971187714909055e-05,
+      "loss": 0.387,
+      "step": 1927
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.971147397022372e-05,
+      "loss": 0.3491,
+      "step": 1928
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9711070511101256e-05,
+      "loss": 0.3906,
+      "step": 1929
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.971066677172772e-05,
+      "loss": 0.4481,
+      "step": 1930
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.971026275210769e-05,
+      "loss": 0.3331,
+      "step": 1931
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.970985845224577e-05,
+      "loss": 0.3215,
+      "step": 1932
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.970945387214652e-05,
+      "loss": 0.2819,
+      "step": 1933
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9709049011814546e-05,
+      "loss": 0.3,
+      "step": 1934
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.970864387125442e-05,
+      "loss": 0.3355,
+      "step": 1935
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9708238450470766e-05,
+      "loss": 0.4343,
+      "step": 1936
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9707832749468155e-05,
+      "loss": 0.3639,
+      "step": 1937
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9707426768251194e-05,
+      "loss": 0.3298,
+      "step": 1938
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.97070205068245e-05,
+      "loss": 0.3366,
+      "step": 1939
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.970661396519267e-05,
+      "loss": 0.4064,
+      "step": 1940
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9706207143360314e-05,
+      "loss": 0.3193,
+      "step": 1941
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.970580004133205e-05,
+      "loss": 0.4563,
+      "step": 1942
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.970539265911249e-05,
+      "loss": 0.3414,
+      "step": 1943
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.970498499670626e-05,
+      "loss": 0.3242,
+      "step": 1944
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.970457705411798e-05,
+      "loss": 0.2716,
+      "step": 1945
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.970416883135227e-05,
+      "loss": 0.3235,
+      "step": 1946
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9703760328413777e-05,
+      "loss": 0.3499,
+      "step": 1947
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.970335154530712e-05,
+      "loss": 0.3482,
+      "step": 1948
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.970294248203694e-05,
+      "loss": 0.3202,
+      "step": 1949
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.970253313860788e-05,
+      "loss": 0.4113,
+      "step": 1950
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.970212351502457e-05,
+      "loss": 0.3331,
+      "step": 1951
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9701713611291665e-05,
+      "loss": 0.3329,
+      "step": 1952
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9701303427413814e-05,
+      "loss": 0.4797,
+      "step": 1953
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.970089296339566e-05,
+      "loss": 0.3526,
+      "step": 1954
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9700482219241875e-05,
+      "loss": 0.3512,
+      "step": 1955
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.97000711949571e-05,
+      "loss": 0.3434,
+      "step": 1956
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9699659890546e-05,
+      "loss": 0.3472,
+      "step": 1957
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.969924830601325e-05,
+      "loss": 0.4372,
+      "step": 1958
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.969883644136351e-05,
+      "loss": 0.2767,
+      "step": 1959
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.969842429660145e-05,
+      "loss": 0.362,
+      "step": 1960
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.969801187173175e-05,
+      "loss": 0.3045,
+      "step": 1961
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9697599166759084e-05,
+      "loss": 0.265,
+      "step": 1962
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9697186181688124e-05,
+      "loss": 0.2845,
+      "step": 1963
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.969677291652357e-05,
+      "loss": 0.3053,
+      "step": 1964
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.969635937127011e-05,
+      "loss": 0.3326,
+      "step": 1965
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.969594554593241e-05,
+      "loss": 0.2649,
+      "step": 1966
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.969553144051518e-05,
+      "loss": 0.3519,
+      "step": 1967
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9695117055023116e-05,
+      "loss": 0.2671,
+      "step": 1968
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.969470238946092e-05,
+      "loss": 0.3394,
+      "step": 1969
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9694287443833285e-05,
+      "loss": 0.2782,
+      "step": 1970
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.969387221814492e-05,
+      "loss": 0.3683,
+      "step": 1971
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.969345671240055e-05,
+      "loss": 0.288,
+      "step": 1972
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.969304092660486e-05,
+      "loss": 0.3499,
+      "step": 1973
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.969262486076259e-05,
+      "loss": 0.2759,
+      "step": 1974
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9692208514878444e-05,
+      "loss": 0.347,
+      "step": 1975
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9691791888957154e-05,
+      "loss": 0.3814,
+      "step": 1976
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.969137498300343e-05,
+      "loss": 0.339,
+      "step": 1977
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.969095779702201e-05,
+      "loss": 0.3655,
+      "step": 1978
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9690540331017634e-05,
+      "loss": 0.3347,
+      "step": 1979
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.969012258499502e-05,
+      "loss": 0.3439,
+      "step": 1980
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9689704558958916e-05,
+      "loss": 0.3219,
+      "step": 1981
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.968928625291406e-05,
+      "loss": 0.3025,
+      "step": 1982
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.968886766686519e-05,
+      "loss": 0.2659,
+      "step": 1983
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.968844880081707e-05,
+      "loss": 0.3727,
+      "step": 1984
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.968802965477443e-05,
+      "loss": 0.3374,
+      "step": 1985
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.968761022874203e-05,
+      "loss": 0.2735,
+      "step": 1986
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.968719052272464e-05,
+      "loss": 0.4567,
+      "step": 1987
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9686770536727006e-05,
+      "loss": 0.3456,
+      "step": 1988
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9686350270753894e-05,
+      "loss": 0.3279,
+      "step": 1989
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9685929724810075e-05,
+      "loss": 0.2878,
+      "step": 1990
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.968550889890031e-05,
+      "loss": 0.3517,
+      "step": 1991
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.968508779302939e-05,
+      "loss": 0.3052,
+      "step": 1992
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.968466640720205e-05,
+      "loss": 0.3309,
+      "step": 1993
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.968424474142311e-05,
+      "loss": 0.3577,
+      "step": 1994
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9683822795697344e-05,
+      "loss": 0.3112,
+      "step": 1995
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9683400570029524e-05,
+      "loss": 0.2303,
+      "step": 1996
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.968297806442446e-05,
+      "loss": 0.4133,
+      "step": 1997
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9682555278886914e-05,
+      "loss": 0.3233,
+      "step": 1998
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9682132213421705e-05,
+      "loss": 0.3256,
+      "step": 1999
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9681708868033616e-05,
+      "loss": 0.303,
+      "step": 2000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.968128524272746e-05,
+      "loss": 0.2975,
+      "step": 2001
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.968086133750803e-05,
+      "loss": 0.2607,
+      "step": 2002
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.968043715238015e-05,
+      "loss": 0.2458,
+      "step": 2003
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9680012687348614e-05,
+      "loss": 0.3388,
+      "step": 2004
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9679587942418245e-05,
+      "loss": 0.2953,
+      "step": 2005
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.967916291759386e-05,
+      "loss": 0.4235,
+      "step": 2006
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.967873761288027e-05,
+      "loss": 0.3281,
+      "step": 2007
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.96783120282823e-05,
+      "loss": 0.4556,
+      "step": 2008
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.96778861638048e-05,
+      "loss": 0.3212,
+      "step": 2009
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.967746001945257e-05,
+      "loss": 0.3614,
+      "step": 2010
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.967703359523045e-05,
+      "loss": 0.4276,
+      "step": 2011
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9676606891143294e-05,
+      "loss": 0.3697,
+      "step": 2012
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.967617990719592e-05,
+      "loss": 0.3003,
+      "step": 2013
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.967575264339318e-05,
+      "loss": 0.3315,
+      "step": 2014
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9675325099739915e-05,
+      "loss": 0.3375,
+      "step": 2015
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9674897276240984e-05,
+      "loss": 0.3686,
+      "step": 2016
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9674469172901226e-05,
+      "loss": 0.3759,
+      "step": 2017
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9674040789725505e-05,
+      "loss": 0.3665,
+      "step": 2018
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9673612126718674e-05,
+      "loss": 0.424,
+      "step": 2019
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.96731831838856e-05,
+      "loss": 0.3703,
+      "step": 2020
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.967275396123114e-05,
+      "loss": 0.3779,
+      "step": 2021
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.967232445876017e-05,
+      "loss": 0.2762,
+      "step": 2022
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.967189467647756e-05,
+      "loss": 0.3006,
+      "step": 2023
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9671464614388174e-05,
+      "loss": 0.2999,
+      "step": 2024
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.96710342724969e-05,
+      "loss": 0.3304,
+      "step": 2025
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9670603650808624e-05,
+      "loss": 0.3742,
+      "step": 2026
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.967017274932821e-05,
+      "loss": 0.3444,
+      "step": 2027
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.966974156806056e-05,
+      "loss": 0.5832,
+      "step": 2028
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.966931010701056e-05,
+      "loss": 0.3784,
+      "step": 2029
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.966887836618311e-05,
+      "loss": 0.3222,
+      "step": 2030
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.96684463455831e-05,
+      "loss": 0.3066,
+      "step": 2031
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9668014045215416e-05,
+      "loss": 0.339,
+      "step": 2032
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.966758146508499e-05,
+      "loss": 0.3633,
+      "step": 2033
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.96671486051967e-05,
+      "loss": 0.2998,
+      "step": 2034
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.966671546555548e-05,
+      "loss": 0.273,
+      "step": 2035
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.966628204616622e-05,
+      "loss": 0.2992,
+      "step": 2036
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.966584834703385e-05,
+      "loss": 0.3826,
+      "step": 2037
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.966541436816328e-05,
+      "loss": 0.4626,
+      "step": 2038
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.966498010955944e-05,
+      "loss": 0.3446,
+      "step": 2039
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.966454557122725e-05,
+      "loss": 0.4245,
+      "step": 2040
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.966411075317163e-05,
+      "loss": 0.3514,
+      "step": 2041
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.966367565539753e-05,
+      "loss": 0.319,
+      "step": 2042
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9663240277909875e-05,
+      "loss": 0.32,
+      "step": 2043
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.96628046207136e-05,
+      "loss": 0.408,
+      "step": 2044
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.966236868381365e-05,
+      "loss": 0.2861,
+      "step": 2045
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9661932467214965e-05,
+      "loss": 0.4212,
+      "step": 2046
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.966149597092249e-05,
+      "loss": 0.3195,
+      "step": 2047
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.966105919494118e-05,
+      "loss": 0.3814,
+      "step": 2048
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9660622139276005e-05,
+      "loss": 0.3389,
+      "step": 2049
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9660184803931885e-05,
+      "loss": 0.2827,
+      "step": 2050
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.965974718891381e-05,
+      "loss": 0.2814,
+      "step": 2051
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.965930929422673e-05,
+      "loss": 0.3643,
+      "step": 2052
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.965887111987561e-05,
+      "loss": 0.4139,
+      "step": 2053
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.965843266586543e-05,
+      "loss": 0.291,
+      "step": 2054
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9657993932201153e-05,
+      "loss": 0.2921,
+      "step": 2055
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.965755491888775e-05,
+      "loss": 0.3062,
+      "step": 2056
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.965711562593022e-05,
+      "loss": 0.3101,
+      "step": 2057
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.965667605333353e-05,
+      "loss": 0.3476,
+      "step": 2058
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.965623620110267e-05,
+      "loss": 0.4434,
+      "step": 2059
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.965579606924262e-05,
+      "loss": 0.4242,
+      "step": 2060
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9655355657758376e-05,
+      "loss": 0.3302,
+      "step": 2061
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.965491496665494e-05,
+      "loss": 0.331,
+      "step": 2062
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9654473995937304e-05,
+      "loss": 0.3436,
+      "step": 2063
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9654032745610466e-05,
+      "loss": 0.3034,
+      "step": 2064
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.965359121567944e-05,
+      "loss": 0.2626,
+      "step": 2065
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.965314940614922e-05,
+      "loss": 0.3663,
+      "step": 2066
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9652707317024834e-05,
+      "loss": 0.3436,
+      "step": 2067
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9652264948311275e-05,
+      "loss": 0.4055,
+      "step": 2068
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9651822300013586e-05,
+      "loss": 0.4192,
+      "step": 2069
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9651379372136756e-05,
+      "loss": 0.4138,
+      "step": 2070
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.965093616468583e-05,
+      "loss": 0.6876,
+      "step": 2071
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.965049267766583e-05,
+      "loss": 0.3842,
+      "step": 2072
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.965004891108178e-05,
+      "loss": 0.3574,
+      "step": 2073
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9649604864938726e-05,
+      "loss": 0.3817,
+      "step": 2074
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.964916053924168e-05,
+      "loss": 0.309,
+      "step": 2075
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.964871593399571e-05,
+      "loss": 0.3914,
+      "step": 2076
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.964827104920584e-05,
+      "loss": 0.4492,
+      "step": 2077
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9647825884877117e-05,
+      "loss": 0.3092,
+      "step": 2078
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.96473804410146e-05,
+      "loss": 0.3803,
+      "step": 2079
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.964693471762333e-05,
+      "loss": 0.9852,
+      "step": 2080
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.964648871470836e-05,
+      "loss": 0.3066,
+      "step": 2081
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.964604243227477e-05,
+      "loss": 0.4216,
+      "step": 2082
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9645595870327584e-05,
+      "loss": 0.3216,
+      "step": 2083
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9645149028871895e-05,
+      "loss": 0.3431,
+      "step": 2084
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.964470190791277e-05,
+      "loss": 0.3649,
+      "step": 2085
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9644254507455266e-05,
+      "loss": 0.3705,
+      "step": 2086
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9643806827504464e-05,
+      "loss": 0.7129,
+      "step": 2087
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9643358868065444e-05,
+      "loss": 0.3177,
+      "step": 2088
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.964291062914328e-05,
+      "loss": 0.3772,
+      "step": 2089
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.964246211074307e-05,
+      "loss": 0.7771,
+      "step": 2090
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.964201331286988e-05,
+      "loss": 0.3448,
+      "step": 2091
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.964156423552881e-05,
+      "loss": 0.5983,
+      "step": 2092
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9641114878724956e-05,
+      "loss": 0.3808,
+      "step": 2093
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.964066524246341e-05,
+      "loss": 0.2967,
+      "step": 2094
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.964021532674927e-05,
+      "loss": 0.6865,
+      "step": 2095
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.963976513158764e-05,
+      "loss": 0.3938,
+      "step": 2096
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9639314656983625e-05,
+      "loss": 0.4377,
+      "step": 2097
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.963886390294235e-05,
+      "loss": 0.6823,
+      "step": 2098
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.96384128694689e-05,
+      "loss": 0.6345,
+      "step": 2099
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9637961556568404e-05,
+      "loss": 0.4166,
+      "step": 2100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9637509964245975e-05,
+      "loss": 0.4378,
+      "step": 2101
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.963705809250675e-05,
+      "loss": 0.9226,
+      "step": 2102
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.963660594135584e-05,
+      "loss": 0.3686,
+      "step": 2103
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.963615351079837e-05,
+      "loss": 0.3889,
+      "step": 2104
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.963570080083947e-05,
+      "loss": 0.6654,
+      "step": 2105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.963524781148429e-05,
+      "loss": 0.302,
+      "step": 2106
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.963479454273796e-05,
+      "loss": 0.3566,
+      "step": 2107
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.963434099460561e-05,
+      "loss": 0.421,
+      "step": 2108
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9633887167092404e-05,
+      "loss": 0.3396,
+      "step": 2109
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.963343306020347e-05,
+      "loss": 0.3294,
+      "step": 2110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9632978673943964e-05,
+      "loss": 0.6285,
+      "step": 2111
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9632524008319045e-05,
+      "loss": 0.6436,
+      "step": 2112
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9632069063333864e-05,
+      "loss": 0.8975,
+      "step": 2113
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.963161383899358e-05,
+      "loss": 0.3036,
+      "step": 2114
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.963115833530336e-05,
+      "loss": 0.3913,
+      "step": 2115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.963070255226836e-05,
+      "loss": 0.3492,
+      "step": 2116
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.963024648989376e-05,
+      "loss": 0.302,
+      "step": 2117
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.962979014818473e-05,
+      "loss": 0.6709,
+      "step": 2118
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.962933352714645e-05,
+      "loss": 0.4331,
+      "step": 2119
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.962887662678408e-05,
+      "loss": 0.7286,
+      "step": 2120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9628419447102824e-05,
+      "loss": 0.6378,
+      "step": 2121
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.962796198810785e-05,
+      "loss": 0.2434,
+      "step": 2122
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9627504249804365e-05,
+      "loss": 0.378,
+      "step": 2123
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.962704623219754e-05,
+      "loss": 0.3387,
+      "step": 2124
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.962658793529258e-05,
+      "loss": 0.6559,
+      "step": 2125
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.962612935909468e-05,
+      "loss": 0.6061,
+      "step": 2126
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.962567050360904e-05,
+      "loss": 0.7081,
+      "step": 2127
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.962521136884087e-05,
+      "loss": 0.4309,
+      "step": 2128
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9624751954795364e-05,
+      "loss": 0.3525,
+      "step": 2129
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.962429226147775e-05,
+      "loss": 0.2984,
+      "step": 2130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.962383228889323e-05,
+      "loss": 0.3864,
+      "step": 2131
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9623372037047026e-05,
+      "loss": 0.7975,
+      "step": 2132
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.962291150594434e-05,
+      "loss": 0.3735,
+      "step": 2133
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.962245069559042e-05,
+      "loss": 0.3824,
+      "step": 2134
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.962198960599048e-05,
+      "loss": 0.3467,
+      "step": 2135
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.962152823714975e-05,
+      "loss": 0.3895,
+      "step": 2136
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.962106658907347e-05,
+      "loss": 0.3436,
+      "step": 2137
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.962060466176687e-05,
+      "loss": 0.3203,
+      "step": 2138
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.962014245523517e-05,
+      "loss": 0.3886,
+      "step": 2139
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9619679969483646e-05,
+      "loss": 0.3428,
+      "step": 2140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.961921720451752e-05,
+      "loss": 0.3611,
+      "step": 2141
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.961875416034205e-05,
+      "loss": 0.3585,
+      "step": 2142
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.961829083696248e-05,
+      "loss": 0.3258,
+      "step": 2143
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.961782723438407e-05,
+      "loss": 0.396,
+      "step": 2144
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.961736335261209e-05,
+      "loss": 0.261,
+      "step": 2145
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9616899191651775e-05,
+      "loss": 0.404,
+      "step": 2146
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.96164347515084e-05,
+      "loss": 0.4012,
+      "step": 2147
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9615970032187244e-05,
+      "loss": 0.2992,
+      "step": 2148
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9615505033693554e-05,
+      "loss": 0.4189,
+      "step": 2149
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9615039756032625e-05,
+      "loss": 0.2802,
+      "step": 2150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.961457419920973e-05,
+      "loss": 0.3688,
+      "step": 2151
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9614108363230135e-05,
+      "loss": 0.3206,
+      "step": 2152
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.961364224809914e-05,
+      "loss": 0.3394,
+      "step": 2153
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9613175853822024e-05,
+      "loss": 0.2714,
+      "step": 2154
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.961270918040407e-05,
+      "loss": 0.3797,
+      "step": 2155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.961224222785058e-05,
+      "loss": 0.3708,
+      "step": 2156
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9611774996166854e-05,
+      "loss": 0.4388,
+      "step": 2157
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.961130748535818e-05,
+      "loss": 0.3519,
+      "step": 2158
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.961083969542986e-05,
+      "loss": 0.3301,
+      "step": 2159
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.961037162638721e-05,
+      "loss": 0.3006,
+      "step": 2160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9609903278235526e-05,
+      "loss": 0.4285,
+      "step": 2161
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9609434650980124e-05,
+      "loss": 0.338,
+      "step": 2162
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.960896574462632e-05,
+      "loss": 0.4099,
+      "step": 2163
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.960849655917944e-05,
+      "loss": 0.4853,
+      "step": 2164
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9608027094644796e-05,
+      "loss": 0.3358,
+      "step": 2165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.960755735102771e-05,
+      "loss": 0.2781,
+      "step": 2166
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.960708732833351e-05,
+      "loss": 0.3263,
+      "step": 2167
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9606617026567534e-05,
+      "loss": 0.3995,
+      "step": 2168
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.960614644573511e-05,
+      "loss": 0.582,
+      "step": 2169
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.960567558584157e-05,
+      "loss": 0.3848,
+      "step": 2170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.960520444689226e-05,
+      "loss": 0.3876,
+      "step": 2171
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.960473302889254e-05,
+      "loss": 0.3028,
+      "step": 2172
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9604261331847724e-05,
+      "loss": 0.3581,
+      "step": 2173
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.960378935576317e-05,
+      "loss": 0.6108,
+      "step": 2174
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.960331710064425e-05,
+      "loss": 0.3429,
+      "step": 2175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.960284456649631e-05,
+      "loss": 0.5047,
+      "step": 2176
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.960237175332471e-05,
+      "loss": 0.3638,
+      "step": 2177
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.96018986611348e-05,
+      "loss": 0.3706,
+      "step": 2178
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9601425289931955e-05,
+      "loss": 0.5007,
+      "step": 2179
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9600951639721546e-05,
+      "loss": 0.3178,
+      "step": 2180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.960047771050894e-05,
+      "loss": 0.3283,
+      "step": 2181
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.960000350229952e-05,
+      "loss": 0.3178,
+      "step": 2182
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.959952901509865e-05,
+      "loss": 0.2871,
+      "step": 2183
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.959905424891172e-05,
+      "loss": 0.2552,
+      "step": 2184
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.959857920374412e-05,
+      "loss": 0.319,
+      "step": 2185
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.959810387960122e-05,
+      "loss": 0.3458,
+      "step": 2186
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.959762827648843e-05,
+      "loss": 0.392,
+      "step": 2187
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9597152394411136e-05,
+      "loss": 0.5004,
+      "step": 2188
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.959667623337473e-05,
+      "loss": 0.3592,
+      "step": 2189
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.959619979338461e-05,
+      "loss": 0.2613,
+      "step": 2190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.95957230744462e-05,
+      "loss": 0.4415,
+      "step": 2191
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.959524607656489e-05,
+      "loss": 0.4071,
+      "step": 2192
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.959476879974608e-05,
+      "loss": 0.3063,
+      "step": 2193
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.959429124399521e-05,
+      "loss": 0.3102,
+      "step": 2194
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.959381340931768e-05,
+      "loss": 0.2855,
+      "step": 2195
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.95933352957189e-05,
+      "loss": 0.2751,
+      "step": 2196
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.959285690320431e-05,
+      "loss": 0.3327,
+      "step": 2197
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.959237823177933e-05,
+      "loss": 0.325,
+      "step": 2198
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.959189928144938e-05,
+      "loss": 0.4196,
+      "step": 2199
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.959142005221991e-05,
+      "loss": 0.3657,
+      "step": 2200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9590940544096326e-05,
+      "loss": 0.3036,
+      "step": 2201
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9590460757084096e-05,
+      "loss": 0.2887,
+      "step": 2202
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.958998069118864e-05,
+      "loss": 0.3171,
+      "step": 2203
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9589500346415425e-05,
+      "loss": 0.2795,
+      "step": 2204
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.958901972276988e-05,
+      "loss": 0.3503,
+      "step": 2205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9588538820257455e-05,
+      "loss": 0.3164,
+      "step": 2206
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9588057638883616e-05,
+      "loss": 0.4417,
+      "step": 2207
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.958757617865382e-05,
+      "loss": 0.3278,
+      "step": 2208
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.95870944395735e-05,
+      "loss": 0.3882,
+      "step": 2209
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.958661242164816e-05,
+      "loss": 0.3268,
+      "step": 2210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.958613012488324e-05,
+      "loss": 0.2597,
+      "step": 2211
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.958564754928423e-05,
+      "loss": 0.2761,
+      "step": 2212
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9585164694856576e-05,
+      "loss": 0.3483,
+      "step": 2213
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.958468156160577e-05,
+      "loss": 0.329,
+      "step": 2214
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.958419814953729e-05,
+      "loss": 0.2979,
+      "step": 2215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.958371445865663e-05,
+      "loss": 0.3559,
+      "step": 2216
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.958323048896924e-05,
+      "loss": 0.3114,
+      "step": 2217
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.958274624048065e-05,
+      "loss": 0.2555,
+      "step": 2218
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.958226171319633e-05,
+      "loss": 0.3062,
+      "step": 2219
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9581776907121784e-05,
+      "loss": 0.4919,
+      "step": 2220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.95812918222625e-05,
+      "loss": 0.2856,
+      "step": 2221
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.958080645862399e-05,
+      "loss": 0.4015,
+      "step": 2222
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.958032081621175e-05,
+      "loss": 0.2473,
+      "step": 2223
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.957983489503129e-05,
+      "loss": 0.3073,
+      "step": 2224
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9579348695088124e-05,
+      "loss": 0.2867,
+      "step": 2225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.957886221638777e-05,
+      "loss": 0.2623,
+      "step": 2226
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.957837545893573e-05,
+      "loss": 0.2776,
+      "step": 2227
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.957788842273754e-05,
+      "loss": 0.3897,
+      "step": 2228
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.957740110779872e-05,
+      "loss": 0.3012,
+      "step": 2229
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.957691351412478e-05,
+      "loss": 0.3036,
+      "step": 2230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.957642564172127e-05,
+      "loss": 0.2792,
+      "step": 2231
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9575937490593726e-05,
+      "loss": 0.3696,
+      "step": 2232
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.957544906074767e-05,
+      "loss": 0.4556,
+      "step": 2233
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9574960352188645e-05,
+      "loss": 0.3732,
+      "step": 2234
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9574471364922195e-05,
+      "loss": 0.2578,
+      "step": 2235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9573982098953866e-05,
+      "loss": 0.3326,
+      "step": 2236
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9573492554289206e-05,
+      "loss": 0.2641,
+      "step": 2237
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.957300273093377e-05,
+      "loss": 0.2649,
+      "step": 2238
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.957251262889311e-05,
+      "loss": 0.2639,
+      "step": 2239
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9572022248172786e-05,
+      "loss": 0.2962,
+      "step": 2240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.957153158877835e-05,
+      "loss": 0.213,
+      "step": 2241
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.957104065071538e-05,
+      "loss": 0.3733,
+      "step": 2242
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.957054943398944e-05,
+      "loss": 0.2983,
+      "step": 2243
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.957005793860611e-05,
+      "loss": 0.3179,
+      "step": 2244
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9569566164570933e-05,
+      "loss": 0.2891,
+      "step": 2245
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9569074111889524e-05,
+      "loss": 0.3933,
+      "step": 2246
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.956858178056744e-05,
+      "loss": 0.2469,
+      "step": 2247
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9568089170610264e-05,
+      "loss": 0.3351,
+      "step": 2248
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9567596282023596e-05,
+      "loss": 0.3356,
+      "step": 2249
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.956710311481303e-05,
+      "loss": 0.3565,
+      "step": 2250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9566609668984134e-05,
+      "loss": 0.3491,
+      "step": 2251
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.956611594454253e-05,
+      "loss": 0.3332,
+      "step": 2252
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.95656219414938e-05,
+      "loss": 0.277,
+      "step": 2253
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.956512765984355e-05,
+      "loss": 0.4199,
+      "step": 2254
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.956463309959739e-05,
+      "loss": 0.2531,
+      "step": 2255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9564138260760935e-05,
+      "loss": 0.4283,
+      "step": 2256
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9563643143339786e-05,
+      "loss": 0.3388,
+      "step": 2257
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9563147747339556e-05,
+      "loss": 0.4038,
+      "step": 2258
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.956265207276588e-05,
+      "loss": 0.3028,
+      "step": 2259
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.956215611962436e-05,
+      "loss": 0.29,
+      "step": 2260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.956165988792063e-05,
+      "loss": 0.2687,
+      "step": 2261
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.956116337766031e-05,
+      "loss": 0.3261,
+      "step": 2262
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.956066658884905e-05,
+      "loss": 0.2798,
+      "step": 2263
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.956016952149247e-05,
+      "loss": 0.3257,
+      "step": 2264
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9559672175596214e-05,
+      "loss": 0.3878,
+      "step": 2265
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.955917455116591e-05,
+      "loss": 0.4051,
+      "step": 2266
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.955867664820721e-05,
+      "loss": 0.4091,
+      "step": 2267
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9558178466725754e-05,
+      "loss": 0.404,
+      "step": 2268
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.955768000672722e-05,
+      "loss": 0.3567,
+      "step": 2269
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9557181268217227e-05,
+      "loss": 0.483,
+      "step": 2270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9556682251201434e-05,
+      "loss": 0.3298,
+      "step": 2271
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.955618295568552e-05,
+      "loss": 0.3105,
+      "step": 2272
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.955568338167513e-05,
+      "loss": 0.2945,
+      "step": 2273
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.955518352917595e-05,
+      "loss": 0.399,
+      "step": 2274
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9554683398193626e-05,
+      "loss": 0.3468,
+      "step": 2275
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9554182988733846e-05,
+      "loss": 0.3032,
+      "step": 2276
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9553682300802274e-05,
+      "loss": 0.3609,
+      "step": 2277
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9553181334404596e-05,
+      "loss": 0.3205,
+      "step": 2278
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9552680089546496e-05,
+      "loss": 0.287,
+      "step": 2279
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.955217856623365e-05,
+      "loss": 0.5212,
+      "step": 2280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.955167676447174e-05,
+      "loss": 0.4467,
+      "step": 2281
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.955117468426649e-05,
+      "loss": 0.3268,
+      "step": 2282
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.955067232562355e-05,
+      "loss": 0.4494,
+      "step": 2283
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9550169688548645e-05,
+      "loss": 0.3649,
+      "step": 2284
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.954966677304747e-05,
+      "loss": 0.2323,
+      "step": 2285
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.954916357912573e-05,
+      "loss": 0.3517,
+      "step": 2286
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.954866010678912e-05,
+      "loss": 0.3436,
+      "step": 2287
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9548156356043375e-05,
+      "loss": 0.3535,
+      "step": 2288
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.954765232689418e-05,
+      "loss": 0.308,
+      "step": 2289
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.954714801934726e-05,
+      "loss": 0.2886,
+      "step": 2290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.954664343340834e-05,
+      "loss": 0.302,
+      "step": 2291
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.954613856908315e-05,
+      "loss": 0.3903,
+      "step": 2292
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9545633426377404e-05,
+      "loss": 0.4832,
+      "step": 2293
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.954512800529683e-05,
+      "loss": 0.3742,
+      "step": 2294
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.954462230584716e-05,
+      "loss": 0.3672,
+      "step": 2295
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.954411632803413e-05,
+      "loss": 0.4002,
+      "step": 2296
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.954361007186349e-05,
+      "loss": 0.3554,
+      "step": 2297
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.954310353734096e-05,
+      "loss": 0.3661,
+      "step": 2298
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9542596724472306e-05,
+      "loss": 0.3218,
+      "step": 2299
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.954208963326327e-05,
+      "loss": 0.3345,
+      "step": 2300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9541582263719586e-05,
+      "loss": 0.34,
+      "step": 2301
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.954107461584703e-05,
+      "loss": 0.3652,
+      "step": 2302
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.954056668965135e-05,
+      "loss": 0.3002,
+      "step": 2303
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9540058485138306e-05,
+      "loss": 0.3179,
+      "step": 2304
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9539550002313666e-05,
+      "loss": 0.3174,
+      "step": 2305
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.953904124118319e-05,
+      "loss": 0.2277,
+      "step": 2306
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9538532201752665e-05,
+      "loss": 0.2044,
+      "step": 2307
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9538022884027835e-05,
+      "loss": 0.4122,
+      "step": 2308
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.95375132880145e-05,
+      "loss": 0.322,
+      "step": 2309
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.953700341371842e-05,
+      "loss": 0.357,
+      "step": 2310
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.95364932611454e-05,
+      "loss": 0.2777,
+      "step": 2311
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.953598283030121e-05,
+      "loss": 0.4798,
+      "step": 2312
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.953547212119165e-05,
+      "loss": 0.2915,
+      "step": 2313
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.95349611338225e-05,
+      "loss": 0.4199,
+      "step": 2314
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.953444986819956e-05,
+      "loss": 0.3355,
+      "step": 2315
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.953393832432863e-05,
+      "loss": 0.3614,
+      "step": 2316
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.953342650221552e-05,
+      "loss": 0.2613,
+      "step": 2317
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.953291440186601e-05,
+      "loss": 0.3709,
+      "step": 2318
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.953240202328593e-05,
+      "loss": 0.3655,
+      "step": 2319
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9531889366481086e-05,
+      "loss": 0.2712,
+      "step": 2320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.953137643145729e-05,
+      "loss": 0.3362,
+      "step": 2321
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9530863218220355e-05,
+      "loss": 0.4126,
+      "step": 2322
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9530349726776105e-05,
+      "loss": 0.338,
+      "step": 2323
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.952983595713037e-05,
+      "loss": 0.3496,
+      "step": 2324
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.952932190928896e-05,
+      "loss": 0.2662,
+      "step": 2325
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9528807583257724e-05,
+      "loss": 0.3612,
+      "step": 2326
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.952829297904249e-05,
+      "loss": 0.2871,
+      "step": 2327
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.952777809664908e-05,
+      "loss": 0.3199,
+      "step": 2328
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.952726293608335e-05,
+      "loss": 0.2042,
+      "step": 2329
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9526747497351146e-05,
+      "loss": 0.2853,
+      "step": 2330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.952623178045829e-05,
+      "loss": 0.3627,
+      "step": 2331
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.952571578541066e-05,
+      "loss": 0.2923,
+      "step": 2332
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.952519951221408e-05,
+      "loss": 0.2489,
+      "step": 2333
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.952468296087443e-05,
+      "loss": 0.3572,
+      "step": 2334
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.952416613139754e-05,
+      "loss": 0.2747,
+      "step": 2335
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9523649023789295e-05,
+      "loss": 0.3579,
+      "step": 2336
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.952313163805556e-05,
+      "loss": 0.3762,
+      "step": 2337
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.952261397420219e-05,
+      "loss": 0.2698,
+      "step": 2338
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.952209603223506e-05,
+      "loss": 0.2987,
+      "step": 2339
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9521577812160046e-05,
+      "loss": 0.3208,
+      "step": 2340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.952105931398303e-05,
+      "loss": 0.3211,
+      "step": 2341
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.952054053770988e-05,
+      "loss": 0.3186,
+      "step": 2342
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.952002148334649e-05,
+      "loss": 0.2225,
+      "step": 2343
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.951950215089874e-05,
+      "loss": 0.334,
+      "step": 2344
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9518982540372525e-05,
+      "loss": 0.3319,
+      "step": 2345
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.951846265177373e-05,
+      "loss": 0.3291,
+      "step": 2346
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.951794248510827e-05,
+      "loss": 0.3155,
+      "step": 2347
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.951742204038202e-05,
+      "loss": 0.3601,
+      "step": 2348
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9516901317600895e-05,
+      "loss": 0.2836,
+      "step": 2349
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9516380316770804e-05,
+      "loss": 0.3917,
+      "step": 2350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.951585903789764e-05,
+      "loss": 0.3066,
+      "step": 2351
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.951533748098734e-05,
+      "loss": 0.359,
+      "step": 2352
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9514815646045795e-05,
+      "loss": 0.3393,
+      "step": 2353
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9514293533078934e-05,
+      "loss": 0.339,
+      "step": 2354
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.951377114209268e-05,
+      "loss": 0.3427,
+      "step": 2355
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9513248473092956e-05,
+      "loss": 0.3466,
+      "step": 2356
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.951272552608569e-05,
+      "loss": 0.3621,
+      "step": 2357
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9512202301076806e-05,
+      "loss": 0.2713,
+      "step": 2358
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.951167879807224e-05,
+      "loss": 0.3155,
+      "step": 2359
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.951115501707794e-05,
+      "loss": 0.6457,
+      "step": 2360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.951063095809983e-05,
+      "loss": 0.2627,
+      "step": 2361
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.951010662114387e-05,
+      "loss": 0.2508,
+      "step": 2362
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9509582006216e-05,
+      "loss": 0.2984,
+      "step": 2363
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9509057113322165e-05,
+      "loss": 0.4234,
+      "step": 2364
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.950853194246832e-05,
+      "loss": 0.2673,
+      "step": 2365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.950800649366042e-05,
+      "loss": 0.2747,
+      "step": 2366
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9507480766904435e-05,
+      "loss": 0.4531,
+      "step": 2367
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.950695476220631e-05,
+      "loss": 0.3441,
+      "step": 2368
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.950642847957202e-05,
+      "loss": 0.3308,
+      "step": 2369
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.950590191900754e-05,
+      "loss": 0.2914,
+      "step": 2370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9505375080518824e-05,
+      "loss": 0.2832,
+      "step": 2371
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.950484796411187e-05,
+      "loss": 0.2048,
+      "step": 2372
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.950432056979263e-05,
+      "loss": 0.2897,
+      "step": 2373
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9503792897567106e-05,
+      "loss": 0.3357,
+      "step": 2374
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9503264947441275e-05,
+      "loss": 0.3686,
+      "step": 2375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.950273671942112e-05,
+      "loss": 0.3111,
+      "step": 2376
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.950220821351265e-05,
+      "loss": 0.3247,
+      "step": 2377
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.950167942972183e-05,
+      "loss": 0.3629,
+      "step": 2378
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.950115036805468e-05,
+      "loss": 0.3161,
+      "step": 2379
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.950062102851718e-05,
+      "loss": 0.3351,
+      "step": 2380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.950009141111536e-05,
+      "loss": 0.251,
+      "step": 2381
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.949956151585521e-05,
+      "loss": 0.3425,
+      "step": 2382
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.949903134274274e-05,
+      "loss": 0.3386,
+      "step": 2383
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.949850089178397e-05,
+      "loss": 0.3312,
+      "step": 2384
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.94979701629849e-05,
+      "loss": 0.3708,
+      "step": 2385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.949743915635157e-05,
+      "loss": 0.2445,
+      "step": 2386
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.949690787188998e-05,
+      "loss": 0.2858,
+      "step": 2387
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.949637630960617e-05,
+      "loss": 0.2905,
+      "step": 2388
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.949584446950617e-05,
+      "loss": 0.3537,
+      "step": 2389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.949531235159601e-05,
+      "loss": 0.286,
+      "step": 2390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.949477995588172e-05,
+      "loss": 0.3138,
+      "step": 2391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9494247282369336e-05,
+      "loss": 0.377,
+      "step": 2392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.94937143310649e-05,
+      "loss": 0.2643,
+      "step": 2393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.949318110197447e-05,
+      "loss": 0.3652,
+      "step": 2394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9492647595104074e-05,
+      "loss": 0.3233,
+      "step": 2395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9492113810459775e-05,
+      "loss": 0.2974,
+      "step": 2396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.949157974804762e-05,
+      "loss": 0.5337,
+      "step": 2397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.949104540787367e-05,
+      "loss": 0.2304,
+      "step": 2398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.949051078994399e-05,
+      "loss": 0.2359,
+      "step": 2399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9489975894264636e-05,
+      "loss": 0.2383,
+      "step": 2400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.948944072084167e-05,
+      "loss": 0.3073,
+      "step": 2401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.948890526968117e-05,
+      "loss": 0.2152,
+      "step": 2402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.94883695407892e-05,
+      "loss": 0.3639,
+      "step": 2403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.948783353417184e-05,
+      "loss": 0.5378,
+      "step": 2404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.948729724983519e-05,
+      "loss": 0.231,
+      "step": 2405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.94867606877853e-05,
+      "loss": 0.2666,
+      "step": 2406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9486223848028255e-05,
+      "loss": 0.4196,
+      "step": 2407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.948568673057017e-05,
+      "loss": 0.1947,
+      "step": 2408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.948514933541711e-05,
+      "loss": 0.2313,
+      "step": 2409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.94846116625752e-05,
+      "loss": 0.4726,
+      "step": 2410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9484073712050504e-05,
+      "loss": 0.2546,
+      "step": 2411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.948353548384915e-05,
+      "loss": 0.2537,
+      "step": 2412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9482996977977224e-05,
+      "loss": 0.2069,
+      "step": 2413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.948245819444084e-05,
+      "loss": 0.3658,
+      "step": 2414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.948191913324611e-05,
+      "loss": 0.2797,
+      "step": 2415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.948137979439915e-05,
+      "loss": 0.2146,
+      "step": 2416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.948084017790607e-05,
+      "loss": 0.3443,
+      "step": 2417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.948030028377299e-05,
+      "loss": 0.3314,
+      "step": 2418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.947976011200604e-05,
+      "loss": 0.3703,
+      "step": 2419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9479219662611334e-05,
+      "loss": 0.2271,
+      "step": 2420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.947867893559502e-05,
+      "loss": 0.3478,
+      "step": 2421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.947813793096321e-05,
+      "loss": 0.3006,
+      "step": 2422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.947759664872206e-05,
+      "loss": 0.334,
+      "step": 2423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.947705508887769e-05,
+      "loss": 0.5733,
+      "step": 2424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9476513251436244e-05,
+      "loss": 0.3386,
+      "step": 2425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9475971136403874e-05,
+      "loss": 0.1926,
+      "step": 2426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.947542874378674e-05,
+      "loss": 0.3798,
+      "step": 2427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.947488607359097e-05,
+      "loss": 0.593,
+      "step": 2428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.947434312582273e-05,
+      "loss": 0.2617,
+      "step": 2429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9473799900488185e-05,
+      "loss": 0.3713,
+      "step": 2430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.947325639759347e-05,
+      "loss": 0.3471,
+      "step": 2431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.947271261714478e-05,
+      "loss": 0.3792,
+      "step": 2432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9472168559148264e-05,
+      "loss": 0.1866,
+      "step": 2433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.94716242236101e-05,
+      "loss": 0.286,
+      "step": 2434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.947107961053645e-05,
+      "loss": 0.2816,
+      "step": 2435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9470534719933505e-05,
+      "loss": 0.4248,
+      "step": 2436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9469989551807436e-05,
+      "loss": 0.3294,
+      "step": 2437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9469444106164424e-05,
+      "loss": 0.2521,
+      "step": 2438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9468898383010665e-05,
+      "loss": 0.3381,
+      "step": 2439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.946835238235234e-05,
+      "loss": 0.4223,
+      "step": 2440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9467806104195634e-05,
+      "loss": 0.2928,
+      "step": 2441
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9467259548546764e-05,
+      "loss": 0.2173,
+      "step": 2442
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.946671271541191e-05,
+      "loss": 0.2049,
+      "step": 2443
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.946616560479729e-05,
+      "loss": 0.2468,
+      "step": 2444
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.946561821670909e-05,
+      "loss": 0.2979,
+      "step": 2445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.946507055115352e-05,
+      "loss": 0.4353,
+      "step": 2446
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9464522608136805e-05,
+      "loss": 0.1805,
+      "step": 2447
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.946397438766515e-05,
+      "loss": 0.3097,
+      "step": 2448
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.946342588974477e-05,
+      "loss": 0.4789,
+      "step": 2449
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.94628771143819e-05,
+      "loss": 0.304,
+      "step": 2450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.946232806158275e-05,
+      "loss": 0.3425,
+      "step": 2451
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.946177873135355e-05,
+      "loss": 0.2216,
+      "step": 2452
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9461229123700535e-05,
+      "loss": 0.6314,
+      "step": 2453
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9460679238629925e-05,
+      "loss": 0.9557,
+      "step": 2454
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9460129076147974e-05,
+      "loss": 1.6197,
+      "step": 2455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9459578636260904e-05,
+      "loss": 0.8192,
+      "step": 2456
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9459027918974975e-05,
+      "loss": 0.6797,
+      "step": 2457
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9458476924296415e-05,
+      "loss": 0.6531,
+      "step": 2458
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.945792565223149e-05,
+      "loss": 1.6649,
+      "step": 2459
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.945737410278644e-05,
+      "loss": 1.381,
+      "step": 2460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9456822275967526e-05,
+      "loss": 2.3212,
+      "step": 2461
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9456270171781007e-05,
+      "loss": 1.946,
+      "step": 2462
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.945571779023313e-05,
+      "loss": 0.6629,
+      "step": 2463
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9455165131330186e-05,
+      "loss": 0.5947,
+      "step": 2464
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9454612195078415e-05,
+      "loss": 0.3099,
+      "step": 2465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.945405898148411e-05,
+      "loss": 1.9372,
+      "step": 2466
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.945350549055353e-05,
+      "loss": 1.3031,
+      "step": 2467
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.945295172229296e-05,
+      "loss": 0.6404,
+      "step": 2468
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.945239767670868e-05,
+      "loss": 1.7814,
+      "step": 2469
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9451843353806974e-05,
+      "loss": 1.9569,
+      "step": 2470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9451288753594126e-05,
+      "loss": 1.6321,
+      "step": 2471
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.945073387607643e-05,
+      "loss": 1.3923,
+      "step": 2472
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.945017872126017e-05,
+      "loss": 0.7565,
+      "step": 2473
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.944962328915165e-05,
+      "loss": 1.6022,
+      "step": 2474
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9449067579757166e-05,
+      "loss": 1.6456,
+      "step": 2475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9448511593083026e-05,
+      "loss": 1.6392,
+      "step": 2476
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.944795532913552e-05,
+      "loss": 2.3216,
+      "step": 2477
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9447398787920976e-05,
+      "loss": 1.0671,
+      "step": 2478
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9446841969445696e-05,
+      "loss": 0.6223,
+      "step": 2479
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9446284873716e-05,
+      "loss": 0.6455,
+      "step": 2480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.944572750073819e-05,
+      "loss": 1.6469,
+      "step": 2481
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.944516985051861e-05,
+      "loss": 1.6707,
+      "step": 2482
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9444611923063574e-05,
+      "loss": 0.7843,
+      "step": 2483
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9444053718379405e-05,
+      "loss": 2.167,
+      "step": 2484
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.944349523647244e-05,
+      "loss": 0.9587,
+      "step": 2485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.944293647734901e-05,
+      "loss": 1.2701,
+      "step": 2486
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.944237744101545e-05,
+      "loss": 0.6472,
+      "step": 2487
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.944181812747811e-05,
+      "loss": 0.9769,
+      "step": 2488
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.944125853674332e-05,
+      "loss": 1.9358,
+      "step": 2489
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.944069866881743e-05,
+      "loss": 1.6416,
+      "step": 2490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.94401385237068e-05,
+      "loss": 1.703,
+      "step": 2491
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.943957810141777e-05,
+      "loss": 1.4576,
+      "step": 2492
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.94390174019567e-05,
+      "loss": 1.4309,
+      "step": 2493
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9438456425329955e-05,
+      "loss": 1.6365,
+      "step": 2494
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9437895171543894e-05,
+      "loss": 2.3005,
+      "step": 2495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9437333640604864e-05,
+      "loss": 2.0824,
+      "step": 2496
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.943677183251926e-05,
+      "loss": 1.602,
+      "step": 2497
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9436209747293446e-05,
+      "loss": 1.2544,
+      "step": 2498
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9435647384933784e-05,
+      "loss": 1.8923,
+      "step": 2499
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9435084745446666e-05,
+      "loss": 2.3105,
+      "step": 2500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.943452182883847e-05,
+      "loss": 0.957,
+      "step": 2501
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9433958635115576e-05,
+      "loss": 1.3294,
+      "step": 2502
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.943339516428438e-05,
+      "loss": 1.8253,
+      "step": 2503
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9432831416351256e-05,
+      "loss": 1.2993,
+      "step": 2504
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.943226739132261e-05,
+      "loss": 1.0617,
+      "step": 2505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.943170308920484e-05,
+      "loss": 1.9716,
+      "step": 2506
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.943113851000434e-05,
+      "loss": 0.335,
+      "step": 2507
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.943057365372751e-05,
+      "loss": 0.9895,
+      "step": 2508
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9430008520380764e-05,
+      "loss": 0.9971,
+      "step": 2509
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9429443109970514e-05,
+      "loss": 1.276,
+      "step": 2510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.942887742250316e-05,
+      "loss": 1.2378,
+      "step": 2511
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.942831145798513e-05,
+      "loss": 0.9806,
+      "step": 2512
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.942774521642283e-05,
+      "loss": 1.2555,
+      "step": 2513
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.942717869782269e-05,
+      "loss": 1.3279,
+      "step": 2514
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.942661190219114e-05,
+      "loss": 1.0024,
+      "step": 2515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.942604482953459e-05,
+      "loss": 0.6249,
+      "step": 2516
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.942547747985949e-05,
+      "loss": 0.9496,
+      "step": 2517
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9424909853172276e-05,
+      "loss": 0.9747,
+      "step": 2518
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9424341949479366e-05,
+      "loss": 0.9688,
+      "step": 2519
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9423773768787214e-05,
+      "loss": 1.4226,
+      "step": 2520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9423205311102264e-05,
+      "loss": 0.7574,
+      "step": 2521
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.942263657643095e-05,
+      "loss": 1.5958,
+      "step": 2522
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.942206756477974e-05,
+      "loss": 1.3107,
+      "step": 2523
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.942149827615508e-05,
+      "loss": 1.3572,
+      "step": 2524
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.942092871056343e-05,
+      "loss": 0.5617,
+      "step": 2525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9420358868011226e-05,
+      "loss": 0.19,
+      "step": 2526
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.941978874850497e-05,
+      "loss": 0.2235,
+      "step": 2527
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.94192183520511e-05,
+      "loss": 0.3463,
+      "step": 2528
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9418647678656096e-05,
+      "loss": 0.2457,
+      "step": 2529
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9418076728326424e-05,
+      "loss": 0.2862,
+      "step": 2530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.941750550106857e-05,
+      "loss": 0.2355,
+      "step": 2531
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9416933996888994e-05,
+      "loss": 0.2479,
+      "step": 2532
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.941636221579419e-05,
+      "loss": 0.4586,
+      "step": 2533
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9415790157790645e-05,
+      "loss": 0.3297,
+      "step": 2534
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.941521782288484e-05,
+      "loss": 0.3526,
+      "step": 2535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.941464521108326e-05,
+      "loss": 0.2297,
+      "step": 2536
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9414072322392424e-05,
+      "loss": 0.3099,
+      "step": 2537
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.94134991568188e-05,
+      "loss": 0.279,
+      "step": 2538
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.941292571436891e-05,
+      "loss": 0.2142,
+      "step": 2539
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.941235199504923e-05,
+      "loss": 0.1951,
+      "step": 2540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9411777998866304e-05,
+      "loss": 0.3193,
+      "step": 2541
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.941120372582662e-05,
+      "loss": 0.3065,
+      "step": 2542
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.941062917593668e-05,
+      "loss": 0.2901,
+      "step": 2543
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9410054349203025e-05,
+      "loss": 0.2424,
+      "step": 2544
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.940947924563216e-05,
+      "loss": 0.2962,
+      "step": 2545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.940890386523062e-05,
+      "loss": 0.2156,
+      "step": 2546
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.940832820800491e-05,
+      "loss": 0.3682,
+      "step": 2547
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9407752273961565e-05,
+      "loss": 0.333,
+      "step": 2548
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.940717606310713e-05,
+      "loss": 0.2839,
+      "step": 2549
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.940659957544813e-05,
+      "loss": 0.2729,
+      "step": 2550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9406022810991093e-05,
+      "loss": 0.2979,
+      "step": 2551
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.940544576974258e-05,
+      "loss": 0.2337,
+      "step": 2552
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.940486845170913e-05,
+      "loss": 0.2228,
+      "step": 2553
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.940429085689728e-05,
+      "loss": 0.3435,
+      "step": 2554
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9403712985313585e-05,
+      "loss": 0.3016,
+      "step": 2555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9403134836964606e-05,
+      "loss": 0.2699,
+      "step": 2556
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9402556411856896e-05,
+      "loss": 0.3319,
+      "step": 2557
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9401977709997005e-05,
+      "loss": 0.2999,
+      "step": 2558
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9401398731391516e-05,
+      "loss": 0.3359,
+      "step": 2559
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.940081947604698e-05,
+      "loss": 0.3439,
+      "step": 2560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.940023994396997e-05,
+      "loss": 0.2866,
+      "step": 2561
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9399660135167056e-05,
+      "loss": 0.2455,
+      "step": 2562
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9399080049644816e-05,
+      "loss": 0.2394,
+      "step": 2563
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.939849968740984e-05,
+      "loss": 0.1952,
+      "step": 2564
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.939791904846869e-05,
+      "loss": 0.2687,
+      "step": 2565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9397338132827966e-05,
+      "loss": 0.2969,
+      "step": 2566
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9396756940494245e-05,
+      "loss": 0.2702,
+      "step": 2567
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.939617547147413e-05,
+      "loss": 0.268,
+      "step": 2568
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9395593725774204e-05,
+      "loss": 0.2912,
+      "step": 2569
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.939501170340107e-05,
+      "loss": 0.4232,
+      "step": 2570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9394429404361334e-05,
+      "loss": 0.3032,
+      "step": 2571
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.939384682866159e-05,
+      "loss": 0.3849,
+      "step": 2572
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.939326397630845e-05,
+      "loss": 0.2425,
+      "step": 2573
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.939268084730853e-05,
+      "loss": 0.3511,
+      "step": 2574
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.939209744166843e-05,
+      "loss": 0.2842,
+      "step": 2575
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9391513759394784e-05,
+      "loss": 0.2208,
+      "step": 2576
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9390929800494193e-05,
+      "loss": 0.2129,
+      "step": 2577
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.93903455649733e-05,
+      "loss": 0.2136,
+      "step": 2578
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.938976105283871e-05,
+      "loss": 0.2914,
+      "step": 2579
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.938917626409706e-05,
+      "loss": 0.5245,
+      "step": 2580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9388591198754986e-05,
+      "loss": 0.303,
+      "step": 2581
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.938800585681912e-05,
+      "loss": 0.2044,
+      "step": 2582
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.93874202382961e-05,
+      "loss": 0.1969,
+      "step": 2583
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9386834343192576e-05,
+      "loss": 0.2965,
+      "step": 2584
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.938624817151518e-05,
+      "loss": 0.3097,
+      "step": 2585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.938566172327057e-05,
+      "loss": 0.275,
+      "step": 2586
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.938507499846539e-05,
+      "loss": 0.2521,
+      "step": 2587
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9384487997106287e-05,
+      "loss": 0.244,
+      "step": 2588
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.938390071919994e-05,
+      "loss": 0.1989,
+      "step": 2589
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9383313164752985e-05,
+      "loss": 0.3665,
+      "step": 2590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9382725333772105e-05,
+      "loss": 0.2815,
+      "step": 2591
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.938213722626396e-05,
+      "loss": 0.2665,
+      "step": 2592
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.938154884223522e-05,
+      "loss": 0.1986,
+      "step": 2593
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9380960181692556e-05,
+      "loss": 0.2587,
+      "step": 2594
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9380371244642644e-05,
+      "loss": 0.3635,
+      "step": 2595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.937978203109216e-05,
+      "loss": 0.2409,
+      "step": 2596
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.93791925410478e-05,
+      "loss": 0.1996,
+      "step": 2597
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9378602774516235e-05,
+      "loss": 0.2595,
+      "step": 2598
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.937801273150415e-05,
+      "loss": 0.3197,
+      "step": 2599
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.937742241201826e-05,
+      "loss": 0.3689,
+      "step": 2600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9376831816065236e-05,
+      "loss": 0.258,
+      "step": 2601
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.937624094365179e-05,
+      "loss": 0.3043,
+      "step": 2602
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.937564979478462e-05,
+      "loss": 0.2985,
+      "step": 2603
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.937505836947043e-05,
+      "loss": 0.2959,
+      "step": 2604
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.937446666771591e-05,
+      "loss": 0.2203,
+      "step": 2605
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.93738746895278e-05,
+      "loss": 0.3353,
+      "step": 2606
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.93732824349128e-05,
+      "loss": 0.2749,
+      "step": 2607
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.937268990387763e-05,
+      "loss": 0.416,
+      "step": 2608
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9372097096429e-05,
+      "loss": 0.3001,
+      "step": 2609
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9371504012573646e-05,
+      "loss": 0.3028,
+      "step": 2610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9370910652318287e-05,
+      "loss": 0.2751,
+      "step": 2611
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.937031701566965e-05,
+      "loss": 0.3044,
+      "step": 2612
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.936972310263448e-05,
+      "loss": 0.2364,
+      "step": 2613
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9369128913219497e-05,
+      "loss": 0.2622,
+      "step": 2614
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9368534447431446e-05,
+      "loss": 0.2339,
+      "step": 2615
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.936793970527707e-05,
+      "loss": 0.2355,
+      "step": 2616
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.936734468676312e-05,
+      "loss": 0.304,
+      "step": 2617
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.936674939189633e-05,
+      "loss": 0.3512,
+      "step": 2618
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9366153820683467e-05,
+      "loss": 0.3029,
+      "step": 2619
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9365557973131274e-05,
+      "loss": 0.295,
+      "step": 2620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.936496184924651e-05,
+      "loss": 0.339,
+      "step": 2621
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.936436544903594e-05,
+      "loss": 0.2428,
+      "step": 2622
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.936376877250632e-05,
+      "loss": 0.3423,
+      "step": 2623
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9363171819664434e-05,
+      "loss": 0.2528,
+      "step": 2624
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.936257459051703e-05,
+      "loss": 0.2555,
+      "step": 2625
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.93619770850709e-05,
+      "loss": 0.3165,
+      "step": 2626
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.936137930333281e-05,
+      "loss": 0.2683,
+      "step": 2627
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.936078124530954e-05,
+      "loss": 0.2852,
+      "step": 2628
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.936018291100788e-05,
+      "loss": 0.2646,
+      "step": 2629
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.93595843004346e-05,
+      "loss": 0.3425,
+      "step": 2630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.935898541359651e-05,
+      "loss": 0.2704,
+      "step": 2631
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9358386250500385e-05,
+      "loss": 0.2973,
+      "step": 2632
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.935778681115303e-05,
+      "loss": 0.3331,
+      "step": 2633
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.935718709556124e-05,
+      "loss": 0.2418,
+      "step": 2634
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.935658710373181e-05,
+      "loss": 0.2587,
+      "step": 2635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.935598683567156e-05,
+      "loss": 0.1638,
+      "step": 2636
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.935538629138728e-05,
+      "loss": 0.2851,
+      "step": 2637
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.93547854708858e-05,
+      "loss": 0.2271,
+      "step": 2638
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.935418437417391e-05,
+      "loss": 0.3179,
+      "step": 2639
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9353583001258455e-05,
+      "loss": 0.2556,
+      "step": 2640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.935298135214623e-05,
+      "loss": 0.2971,
+      "step": 2641
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9352379426844074e-05,
+      "loss": 0.2608,
+      "step": 2642
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.935177722535881e-05,
+      "loss": 0.2273,
+      "step": 2643
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.935117474769726e-05,
+      "loss": 0.304,
+      "step": 2644
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.935057199386627e-05,
+      "loss": 0.2926,
+      "step": 2645
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9349968963872664e-05,
+      "loss": 0.2408,
+      "step": 2646
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.934936565772328e-05,
+      "loss": 0.2309,
+      "step": 2647
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.934876207542498e-05,
+      "loss": 0.3569,
+      "step": 2648
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.934815821698459e-05,
+      "loss": 0.2769,
+      "step": 2649
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9347554082408956e-05,
+      "loss": 0.2405,
+      "step": 2650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.934694967170495e-05,
+      "loss": 0.2659,
+      "step": 2651
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.93463449848794e-05,
+      "loss": 0.2915,
+      "step": 2652
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.934574002193919e-05,
+      "loss": 0.2609,
+      "step": 2653
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.934513478289117e-05,
+      "loss": 0.3528,
+      "step": 2654
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.934452926774219e-05,
+      "loss": 0.369,
+      "step": 2655
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.934392347649914e-05,
+      "loss": 0.2942,
+      "step": 2656
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.934331740916888e-05,
+      "loss": 0.2545,
+      "step": 2657
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9342711065758286e-05,
+      "loss": 0.2436,
+      "step": 2658
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9342104446274224e-05,
+      "loss": 0.2264,
+      "step": 2659
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.934149755072359e-05,
+      "loss": 0.2908,
+      "step": 2660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.934089037911326e-05,
+      "loss": 0.2762,
+      "step": 2661
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.934028293145011e-05,
+      "loss": 0.2375,
+      "step": 2662
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.933967520774104e-05,
+      "loss": 0.2551,
+      "step": 2663
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.933906720799295e-05,
+      "loss": 0.2338,
+      "step": 2664
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9338458932212715e-05,
+      "loss": 0.2287,
+      "step": 2665
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.933785038040726e-05,
+      "loss": 0.2666,
+      "step": 2666
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9337241552583456e-05,
+      "loss": 0.2098,
+      "step": 2667
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9336632448748235e-05,
+      "loss": 0.3126,
+      "step": 2668
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.933602306890848e-05,
+      "loss": 0.2444,
+      "step": 2669
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.933541341307112e-05,
+      "loss": 0.2631,
+      "step": 2670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.933480348124307e-05,
+      "loss": 0.1818,
+      "step": 2671
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9334193273431235e-05,
+      "loss": 0.2268,
+      "step": 2672
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9333582789642544e-05,
+      "loss": 0.2821,
+      "step": 2673
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.933297202988392e-05,
+      "loss": 0.2339,
+      "step": 2674
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9332360994162285e-05,
+      "loss": 0.1827,
+      "step": 2675
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9331749682484574e-05,
+      "loss": 0.2371,
+      "step": 2676
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.933113809485771e-05,
+      "loss": 0.2463,
+      "step": 2677
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.933052623128865e-05,
+      "loss": 0.2025,
+      "step": 2678
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9329914091784316e-05,
+      "loss": 0.2305,
+      "step": 2679
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9329301676351646e-05,
+      "loss": 0.248,
+      "step": 2680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.932868898499761e-05,
+      "loss": 0.2344,
+      "step": 2681
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.932807601772913e-05,
+      "loss": 0.2503,
+      "step": 2682
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9327462774553166e-05,
+      "loss": 0.2135,
+      "step": 2683
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.932684925547668e-05,
+      "loss": 0.2123,
+      "step": 2684
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.932623546050662e-05,
+      "loss": 0.2164,
+      "step": 2685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.932562138964995e-05,
+      "loss": 0.2113,
+      "step": 2686
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.932500704291365e-05,
+      "loss": 0.223,
+      "step": 2687
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.932439242030467e-05,
+      "loss": 0.2045,
+      "step": 2688
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.932377752182998e-05,
+      "loss": 0.224,
+      "step": 2689
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9323162347496566e-05,
+      "loss": 0.2549,
+      "step": 2690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9322546897311386e-05,
+      "loss": 0.271,
+      "step": 2691
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.932193117128143e-05,
+      "loss": 0.1878,
+      "step": 2692
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.93213151694137e-05,
+      "loss": 0.1908,
+      "step": 2693
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9320698891715146e-05,
+      "loss": 0.1854,
+      "step": 2694
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.932008233819278e-05,
+      "loss": 0.2443,
+      "step": 2695
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9319465508853585e-05,
+      "loss": 0.2254,
+      "step": 2696
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.931884840370457e-05,
+      "loss": 0.1934,
+      "step": 2697
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.931823102275272e-05,
+      "loss": 0.2813,
+      "step": 2698
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9317613366005036e-05,
+      "loss": 0.2142,
+      "step": 2699
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.931699543346854e-05,
+      "loss": 0.1865,
+      "step": 2700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9316377225150226e-05,
+      "loss": 0.18,
+      "step": 2701
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.93157587410571e-05,
+      "loss": 0.2465,
+      "step": 2702
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.931513998119619e-05,
+      "loss": 0.2271,
+      "step": 2703
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.93145209455745e-05,
+      "loss": 0.2076,
+      "step": 2704
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9313901634199065e-05,
+      "loss": 0.2458,
+      "step": 2705
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9313282047076904e-05,
+      "loss": 0.2455,
+      "step": 2706
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.931266218421504e-05,
+      "loss": 0.2713,
+      "step": 2707
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9312042045620496e-05,
+      "loss": 0.2434,
+      "step": 2708
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9311421631300324e-05,
+      "loss": 0.2384,
+      "step": 2709
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.931080094126155e-05,
+      "loss": 0.2785,
+      "step": 2710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9310179975511206e-05,
+      "loss": 0.2753,
+      "step": 2711
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.930955873405634e-05,
+      "loss": 0.253,
+      "step": 2712
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9308937216904e-05,
+      "loss": 0.2157,
+      "step": 2713
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.930831542406124e-05,
+      "loss": 0.1947,
+      "step": 2714
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.93076933555351e-05,
+      "loss": 0.1821,
+      "step": 2715
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9307071011332645e-05,
+      "loss": 0.1662,
+      "step": 2716
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.930644839146092e-05,
+      "loss": 0.2337,
+      "step": 2717
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9305825495927e-05,
+      "loss": 0.3247,
+      "step": 2718
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9305202324737953e-05,
+      "loss": 0.2631,
+      "step": 2719
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.930457887790083e-05,
+      "loss": 0.2722,
+      "step": 2720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.930395515542271e-05,
+      "loss": 0.239,
+      "step": 2721
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.930333115731066e-05,
+      "loss": 0.2198,
+      "step": 2722
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.930270688357177e-05,
+      "loss": 0.2324,
+      "step": 2723
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9302082334213107e-05,
+      "loss": 0.2783,
+      "step": 2724
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.930145750924177e-05,
+      "loss": 0.2375,
+      "step": 2725
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.930083240866482e-05,
+      "loss": 0.4408,
+      "step": 2726
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9300207032489363e-05,
+      "loss": 0.251,
+      "step": 2727
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9299581380722505e-05,
+      "loss": 0.2719,
+      "step": 2728
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.929895545337131e-05,
+      "loss": 0.2504,
+      "step": 2729
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.92983292504429e-05,
+      "loss": 0.2471,
+      "step": 2730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.929770277194437e-05,
+      "loss": 0.2443,
+      "step": 2731
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9297076017882824e-05,
+      "loss": 0.2825,
+      "step": 2732
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9296448988265373e-05,
+      "loss": 0.2423,
+      "step": 2733
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.929582168309912e-05,
+      "loss": 0.2182,
+      "step": 2734
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.929519410239119e-05,
+      "loss": 0.2205,
+      "step": 2735
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.929456624614869e-05,
+      "loss": 0.1945,
+      "step": 2736
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.929393811437876e-05,
+      "loss": 0.2168,
+      "step": 2737
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.92933097070885e-05,
+      "loss": 0.2551,
+      "step": 2738
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.929268102428505e-05,
+      "loss": 0.2297,
+      "step": 2739
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9292052065975536e-05,
+      "loss": 0.2596,
+      "step": 2740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9291422832167086e-05,
+      "loss": 0.2239,
+      "step": 2741
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.929079332286685e-05,
+      "loss": 0.1985,
+      "step": 2742
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.929016353808196e-05,
+      "loss": 0.2624,
+      "step": 2743
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.928953347781955e-05,
+      "loss": 0.1995,
+      "step": 2744
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9288903142086785e-05,
+      "loss": 0.2102,
+      "step": 2745
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.92882725308908e-05,
+      "loss": 0.2793,
+      "step": 2746
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.928764164423874e-05,
+      "loss": 0.2701,
+      "step": 2747
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.928701048213777e-05,
+      "loss": 0.2237,
+      "step": 2748
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.928637904459506e-05,
+      "loss": 0.2424,
+      "step": 2749
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9285747331617746e-05,
+      "loss": 0.2948,
+      "step": 2750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9285115343213015e-05,
+      "loss": 0.2017,
+      "step": 2751
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.928448307938802e-05,
+      "loss": 0.2935,
+      "step": 2752
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9283850540149936e-05,
+      "loss": 0.254,
+      "step": 2753
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9283217725505935e-05,
+      "loss": 0.1988,
+      "step": 2754
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.928258463546319e-05,
+      "loss": 0.2859,
+      "step": 2755
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.92819512700289e-05,
+      "loss": 0.2423,
+      "step": 2756
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9281317629210224e-05,
+      "loss": 0.263,
+      "step": 2757
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.928068371301436e-05,
+      "loss": 0.2078,
+      "step": 2758
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.92800495214485e-05,
+      "loss": 0.1885,
+      "step": 2759
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.927941505451983e-05,
+      "loss": 0.1928,
+      "step": 2760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.927878031223555e-05,
+      "loss": 0.2969,
+      "step": 2761
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.927814529460285e-05,
+      "loss": 0.2033,
+      "step": 2762
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9277510001628944e-05,
+      "loss": 0.1925,
+      "step": 2763
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.927687443332103e-05,
+      "loss": 0.2291,
+      "step": 2764
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.927623858968631e-05,
+      "loss": 0.314,
+      "step": 2765
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.927560247073201e-05,
+      "loss": 0.3166,
+      "step": 2766
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9274966076465336e-05,
+      "loss": 0.2359,
+      "step": 2767
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.92743294068935e-05,
+      "loss": 0.2188,
+      "step": 2768
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9273692462023744e-05,
+      "loss": 0.2363,
+      "step": 2769
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.927305524186326e-05,
+      "loss": 0.2181,
+      "step": 2770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.927241774641931e-05,
+      "loss": 0.2291,
+      "step": 2771
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.927177997569909e-05,
+      "loss": 0.2047,
+      "step": 2772
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.927114192970985e-05,
+      "loss": 0.2165,
+      "step": 2773
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.927050360845883e-05,
+      "loss": 0.3087,
+      "step": 2774
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.926986501195326e-05,
+      "loss": 0.2662,
+      "step": 2775
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.926922614020039e-05,
+      "loss": 0.2631,
+      "step": 2776
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9268586993207455e-05,
+      "loss": 0.206,
+      "step": 2777
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.926794757098171e-05,
+      "loss": 0.2714,
+      "step": 2778
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.926730787353041e-05,
+      "loss": 0.2901,
+      "step": 2779
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.926666790086081e-05,
+      "loss": 0.2599,
+      "step": 2780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.926602765298016e-05,
+      "loss": 0.6293,
+      "step": 2781
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.926538712989572e-05,
+      "loss": 0.3123,
+      "step": 2782
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9264746331614776e-05,
+      "loss": 0.278,
+      "step": 2783
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.926410525814457e-05,
+      "loss": 0.5939,
+      "step": 2784
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.926346390949238e-05,
+      "loss": 0.2671,
+      "step": 2785
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9262822285665486e-05,
+      "loss": 0.5937,
+      "step": 2786
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.926218038667115e-05,
+      "loss": 0.8578,
+      "step": 2787
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.926153821251668e-05,
+      "loss": 0.2255,
+      "step": 2788
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.926089576320933e-05,
+      "loss": 0.5427,
+      "step": 2789
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.92602530387564e-05,
+      "loss": 0.8115,
+      "step": 2790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.925961003916517e-05,
+      "loss": 0.2649,
+      "step": 2791
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.925896676444295e-05,
+      "loss": 0.2267,
+      "step": 2792
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9258323214597015e-05,
+      "loss": 0.3851,
+      "step": 2793
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.925767938963468e-05,
+      "loss": 0.2795,
+      "step": 2794
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9257035289563236e-05,
+      "loss": 0.5926,
+      "step": 2795
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9256390914389994e-05,
+      "loss": 0.3367,
+      "step": 2796
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.925574626412226e-05,
+      "loss": 0.2576,
+      "step": 2797
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.925510133876734e-05,
+      "loss": 0.2345,
+      "step": 2798
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.925445613833255e-05,
+      "loss": 0.8535,
+      "step": 2799
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9253810662825216e-05,
+      "loss": 0.2467,
+      "step": 2800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.925316491225265e-05,
+      "loss": 0.2612,
+      "step": 2801
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.925251888662218e-05,
+      "loss": 0.248,
+      "step": 2802
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9251872585941125e-05,
+      "loss": 0.2662,
+      "step": 2803
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.925122601021682e-05,
+      "loss": 0.6207,
+      "step": 2804
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9250579159456603e-05,
+      "loss": 0.2634,
+      "step": 2805
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.92499320336678e-05,
+      "loss": 0.4992,
+      "step": 2806
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.924928463285776e-05,
+      "loss": 0.2845,
+      "step": 2807
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.924863695703382e-05,
+      "loss": 0.2501,
+      "step": 2808
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.924798900620332e-05,
+      "loss": 0.5947,
+      "step": 2809
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9247340780373624e-05,
+      "loss": 0.2523,
+      "step": 2810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9246692279552066e-05,
+      "loss": 0.2362,
+      "step": 2811
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.924604350374601e-05,
+      "loss": 0.2505,
+      "step": 2812
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.924539445296282e-05,
+      "loss": 0.3653,
+      "step": 2813
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9244745127209843e-05,
+      "loss": 0.216,
+      "step": 2814
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.924409552649445e-05,
+      "loss": 0.2444,
+      "step": 2815
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9243445650824006e-05,
+      "loss": 0.2178,
+      "step": 2816
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.924279550020588e-05,
+      "loss": 0.2775,
+      "step": 2817
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.924214507464745e-05,
+      "loss": 0.5799,
+      "step": 2818
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.92414943741561e-05,
+      "loss": 0.2684,
+      "step": 2819
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.924084339873919e-05,
+      "loss": 0.2911,
+      "step": 2820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.924019214840412e-05,
+      "loss": 0.198,
+      "step": 2821
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.923954062315827e-05,
+      "loss": 0.2519,
+      "step": 2822
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.923888882300902e-05,
+      "loss": 0.2652,
+      "step": 2823
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.923823674796378e-05,
+      "loss": 0.2671,
+      "step": 2824
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.923758439802993e-05,
+      "loss": 0.5753,
+      "step": 2825
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.923693177321488e-05,
+      "loss": 0.4987,
+      "step": 2826
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.923627887352602e-05,
+      "loss": 0.3534,
+      "step": 2827
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9235625698970756e-05,
+      "loss": 0.2828,
+      "step": 2828
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.923497224955651e-05,
+      "loss": 0.251,
+      "step": 2829
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.923431852529068e-05,
+      "loss": 0.62,
+      "step": 2830
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.923366452618068e-05,
+      "loss": 0.2739,
+      "step": 2831
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9233010252233925e-05,
+      "loss": 0.2394,
+      "step": 2832
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.923235570345785e-05,
+      "loss": 0.2739,
+      "step": 2833
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.923170087985986e-05,
+      "loss": 0.2817,
+      "step": 2834
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9231045781447396e-05,
+      "loss": 0.2661,
+      "step": 2835
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9230390408227876e-05,
+      "loss": 0.9308,
+      "step": 2836
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.922973476020874e-05,
+      "loss": 0.2166,
+      "step": 2837
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.922907883739743e-05,
+      "loss": 0.6271,
+      "step": 2838
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.922842263980137e-05,
+      "loss": 0.2444,
+      "step": 2839
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9227766167428005e-05,
+      "loss": 0.2285,
+      "step": 2840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.922710942028479e-05,
+      "loss": 0.2972,
+      "step": 2841
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.922645239837916e-05,
+      "loss": 0.2636,
+      "step": 2842
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.922579510171858e-05,
+      "loss": 0.2003,
+      "step": 2843
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9225137530310495e-05,
+      "loss": 0.5332,
+      "step": 2844
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.922447968416237e-05,
+      "loss": 0.5235,
+      "step": 2845
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.922382156328165e-05,
+      "loss": 0.4353,
+      "step": 2846
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.922316316767582e-05,
+      "loss": 0.2486,
+      "step": 2847
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.922250449735233e-05,
+      "loss": 0.2904,
+      "step": 2848
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.922184555231867e-05,
+      "loss": 0.2597,
+      "step": 2849
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.922118633258229e-05,
+      "loss": 0.3929,
+      "step": 2850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9220526838150674e-05,
+      "loss": 0.2903,
+      "step": 2851
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9219867069031304e-05,
+      "loss": 0.2468,
+      "step": 2852
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9219207025231665e-05,
+      "loss": 0.3268,
+      "step": 2853
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9218546706759245e-05,
+      "loss": 0.2295,
+      "step": 2854
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9217886113621525e-05,
+      "loss": 0.2645,
+      "step": 2855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.921722524582599e-05,
+      "loss": 0.2784,
+      "step": 2856
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.921656410338016e-05,
+      "loss": 0.2301,
+      "step": 2857
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.921590268629151e-05,
+      "loss": 0.2214,
+      "step": 2858
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9215240994567546e-05,
+      "loss": 0.3016,
+      "step": 2859
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9214579028215776e-05,
+      "loss": 0.2363,
+      "step": 2860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9213916787243714e-05,
+      "loss": 0.2167,
+      "step": 2861
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.921325427165886e-05,
+      "loss": 0.2584,
+      "step": 2862
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9212591481468725e-05,
+      "loss": 0.3394,
+      "step": 2863
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.921192841668084e-05,
+      "loss": 0.4271,
+      "step": 2864
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.921126507730271e-05,
+      "loss": 0.2264,
+      "step": 2865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9210601463341864e-05,
+      "loss": 0.2449,
+      "step": 2866
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9209937574805834e-05,
+      "loss": 0.3812,
+      "step": 2867
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.920927341170214e-05,
+      "loss": 0.3483,
+      "step": 2868
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.920860897403832e-05,
+      "loss": 0.2341,
+      "step": 2869
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.920794426182191e-05,
+      "loss": 0.5021,
+      "step": 2870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.920727927506044e-05,
+      "loss": 0.2876,
+      "step": 2871
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.920661401376146e-05,
+      "loss": 0.2684,
+      "step": 2872
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.920594847793252e-05,
+      "loss": 0.2724,
+      "step": 2873
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9205282667581156e-05,
+      "loss": 0.2282,
+      "step": 2874
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.920461658271492e-05,
+      "loss": 0.2544,
+      "step": 2875
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9203950223341373e-05,
+      "loss": 0.2427,
+      "step": 2876
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.920328358946806e-05,
+      "loss": 0.2271,
+      "step": 2877
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.920261668110257e-05,
+      "loss": 0.2276,
+      "step": 2878
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.920194949825243e-05,
+      "loss": 0.2238,
+      "step": 2879
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9201282040925234e-05,
+      "loss": 0.2149,
+      "step": 2880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.920061430912853e-05,
+      "loss": 0.2784,
+      "step": 2881
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9199946302869914e-05,
+      "loss": 0.2804,
+      "step": 2882
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.919927802215695e-05,
+      "loss": 0.2083,
+      "step": 2883
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.91986094669972e-05,
+      "loss": 0.3795,
+      "step": 2884
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9197940637398284e-05,
+      "loss": 0.2423,
+      "step": 2885
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9197271533367754e-05,
+      "loss": 0.2395,
+      "step": 2886
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.919660215491322e-05,
+      "loss": 0.2394,
+      "step": 2887
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.919593250204225e-05,
+      "loss": 0.2204,
+      "step": 2888
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.919526257476247e-05,
+      "loss": 0.4009,
+      "step": 2889
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.919459237308145e-05,
+      "loss": 0.3292,
+      "step": 2890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.919392189700681e-05,
+      "loss": 0.3505,
+      "step": 2891
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.919325114654614e-05,
+      "loss": 0.2316,
+      "step": 2892
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9192580121707046e-05,
+      "loss": 0.2015,
+      "step": 2893
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9191908822497155e-05,
+      "loss": 0.3236,
+      "step": 2894
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.919123724892406e-05,
+      "loss": 0.265,
+      "step": 2895
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9190565400995406e-05,
+      "loss": 0.3579,
+      "step": 2896
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.918989327871878e-05,
+      "loss": 0.2315,
+      "step": 2897
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9189220882101813e-05,
+      "loss": 0.2549,
+      "step": 2898
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9188548211152144e-05,
+      "loss": 0.2156,
+      "step": 2899
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.918787526587739e-05,
+      "loss": 0.2361,
+      "step": 2900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9187202046285195e-05,
+      "loss": 0.296,
+      "step": 2901
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.918652855238318e-05,
+      "loss": 0.3128,
+      "step": 2902
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9185854784178996e-05,
+      "loss": 0.2209,
+      "step": 2903
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.918518074168027e-05,
+      "loss": 0.1891,
+      "step": 2904
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9184506424894664e-05,
+      "loss": 0.2661,
+      "step": 2905
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9183831833829804e-05,
+      "loss": 0.2628,
+      "step": 2906
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.918315696849335e-05,
+      "loss": 0.2807,
+      "step": 2907
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.918248182889297e-05,
+      "loss": 0.2682,
+      "step": 2908
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9181806415036305e-05,
+      "loss": 0.2094,
+      "step": 2909
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.918113072693101e-05,
+      "loss": 0.2273,
+      "step": 2910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.918045476458477e-05,
+      "loss": 0.2194,
+      "step": 2911
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9179778528005224e-05,
+      "loss": 0.2589,
+      "step": 2912
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9179102017200065e-05,
+      "loss": 0.1983,
+      "step": 2913
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.917842523217695e-05,
+      "loss": 0.2012,
+      "step": 2914
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.917774817294356e-05,
+      "loss": 0.2483,
+      "step": 2915
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.917707083950758e-05,
+      "loss": 0.2447,
+      "step": 2916
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9176393231876685e-05,
+      "loss": 0.2262,
+      "step": 2917
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.917571535005855e-05,
+      "loss": 0.2536,
+      "step": 2918
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.917503719406088e-05,
+      "loss": 0.2044,
+      "step": 2919
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.917435876389136e-05,
+      "loss": 0.2299,
+      "step": 2920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9173680059557684e-05,
+      "loss": 0.2063,
+      "step": 2921
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.917300108106754e-05,
+      "loss": 0.1609,
+      "step": 2922
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.917232182842865e-05,
+      "loss": 0.4448,
+      "step": 2923
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.917164230164869e-05,
+      "loss": 0.2636,
+      "step": 2924
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.917096250073538e-05,
+      "loss": 0.2168,
+      "step": 2925
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.917028242569645e-05,
+      "loss": 0.2315,
+      "step": 2926
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.916960207653958e-05,
+      "loss": 0.2108,
+      "step": 2927
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.91689214532725e-05,
+      "loss": 0.251,
+      "step": 2928
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9168240555902937e-05,
+      "loss": 0.222,
+      "step": 2929
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.916755938443859e-05,
+      "loss": 0.263,
+      "step": 2930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.916687793888721e-05,
+      "loss": 0.2758,
+      "step": 2931
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.916619621925651e-05,
+      "loss": 0.2096,
+      "step": 2932
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.916551422555422e-05,
+      "loss": 0.3605,
+      "step": 2933
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.916483195778809e-05,
+      "loss": 0.2538,
+      "step": 2934
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9164149415965846e-05,
+      "loss": 0.1763,
+      "step": 2935
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.916346660009523e-05,
+      "loss": 0.2397,
+      "step": 2936
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.916278351018398e-05,
+      "loss": 0.2113,
+      "step": 2937
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9162100146239854e-05,
+      "loss": 0.2303,
+      "step": 2938
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9161416508270594e-05,
+      "loss": 0.2022,
+      "step": 2939
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.916073259628397e-05,
+      "loss": 0.2625,
+      "step": 2940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.916004841028771e-05,
+      "loss": 0.2105,
+      "step": 2941
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.915936395028959e-05,
+      "loss": 0.2375,
+      "step": 2942
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.915867921629738e-05,
+      "loss": 0.2324,
+      "step": 2943
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.915799420831884e-05,
+      "loss": 0.2273,
+      "step": 2944
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.915730892636172e-05,
+      "loss": 0.2276,
+      "step": 2945
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.915662337043382e-05,
+      "loss": 0.2012,
+      "step": 2946
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9155937540542894e-05,
+      "loss": 0.2321,
+      "step": 2947
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.915525143669674e-05,
+      "loss": 0.2351,
+      "step": 2948
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9154565058903115e-05,
+      "loss": 0.3021,
+      "step": 2949
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.915387840716982e-05,
+      "loss": 0.2235,
+      "step": 2950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.915319148150464e-05,
+      "loss": 0.2524,
+      "step": 2951
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9152504281915365e-05,
+      "loss": 0.3044,
+      "step": 2952
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.915181680840978e-05,
+      "loss": 0.2031,
+      "step": 2953
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.915112906099569e-05,
+      "loss": 0.2028,
+      "step": 2954
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.915044103968091e-05,
+      "loss": 0.1975,
+      "step": 2955
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.914975274447321e-05,
+      "loss": 0.2412,
+      "step": 2956
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.914906417538042e-05,
+      "loss": 0.1622,
+      "step": 2957
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.914837533241034e-05,
+      "loss": 0.2302,
+      "step": 2958
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9147686215570785e-05,
+      "loss": 0.3003,
+      "step": 2959
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.914699682486957e-05,
+      "loss": 0.259,
+      "step": 2960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.914630716031451e-05,
+      "loss": 0.1934,
+      "step": 2961
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.914561722191343e-05,
+      "loss": 0.2011,
+      "step": 2962
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.914492700967416e-05,
+      "loss": 0.1608,
+      "step": 2963
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.914423652360452e-05,
+      "loss": 0.2314,
+      "step": 2964
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.914354576371234e-05,
+      "loss": 0.1648,
+      "step": 2965
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.914285473000545e-05,
+      "loss": 0.2863,
+      "step": 2966
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.91421634224917e-05,
+      "loss": 0.1938,
+      "step": 2967
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.914147184117893e-05,
+      "loss": 0.2074,
+      "step": 2968
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.914077998607498e-05,
+      "loss": 0.2332,
+      "step": 2969
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.914008785718769e-05,
+      "loss": 0.2547,
+      "step": 2970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9139395454524905e-05,
+      "loss": 0.2046,
+      "step": 2971
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.91387027780945e-05,
+      "loss": 0.2922,
+      "step": 2972
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.913800982790431e-05,
+      "loss": 0.3235,
+      "step": 2973
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.91373166039622e-05,
+      "loss": 0.2197,
+      "step": 2974
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9136623106276034e-05,
+      "loss": 0.1864,
+      "step": 2975
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.913592933485367e-05,
+      "loss": 0.2843,
+      "step": 2976
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.913523528970299e-05,
+      "loss": 0.2935,
+      "step": 2977
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.913454097083185e-05,
+      "loss": 0.2508,
+      "step": 2978
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.913384637824813e-05,
+      "loss": 0.2118,
+      "step": 2979
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9133151511959715e-05,
+      "loss": 0.407,
+      "step": 2980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9132456371974475e-05,
+      "loss": 0.2255,
+      "step": 2981
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.913176095830031e-05,
+      "loss": 0.1955,
+      "step": 2982
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.913106527094508e-05,
+      "loss": 0.2251,
+      "step": 2983
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.91303693099167e-05,
+      "loss": 0.2249,
+      "step": 2984
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.912967307522305e-05,
+      "loss": 0.2858,
+      "step": 2985
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.912897656687202e-05,
+      "loss": 0.2266,
+      "step": 2986
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.912827978487152e-05,
+      "loss": 0.2461,
+      "step": 2987
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.912758272922946e-05,
+      "loss": 0.2664,
+      "step": 2988
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.912688539995373e-05,
+      "loss": 0.2539,
+      "step": 2989
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9126187797052245e-05,
+      "loss": 0.4307,
+      "step": 2990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9125489920532916e-05,
+      "loss": 0.2132,
+      "step": 2991
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9124791770403655e-05,
+      "loss": 0.261,
+      "step": 2992
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9124093346672385e-05,
+      "loss": 0.2291,
+      "step": 2993
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.912339464934702e-05,
+      "loss": 0.2164,
+      "step": 2994
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9122695678435494e-05,
+      "loss": 0.2151,
+      "step": 2995
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.912199643394573e-05,
+      "loss": 0.2536,
+      "step": 2996
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.912129691588564e-05,
+      "loss": 0.2413,
+      "step": 2997
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.912059712426319e-05,
+      "loss": 0.2754,
+      "step": 2998
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9119897059086297e-05,
+      "loss": 0.2596,
+      "step": 2999
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.91191967203629e-05,
+      "loss": 0.2062,
+      "step": 3000
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 0.43695369362831116,
+      "eval_runtime": 11.5057,
+      "eval_samples_per_second": 2.868,
+      "eval_steps_per_second": 1.478,
+      "step": 3000
+    }
+  ],
+  "max_steps": 30000,
+  "num_train_epochs": 9223372036854775807,
+  "total_flos": 6.01482923606016e+17,
+  "trial_name": null,
+  "trial_params": null
+}