diff --git "a/checkpoint-1000/trainer_state.json" "b/checkpoint-1000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1000/trainer_state.json"
@@ -0,0 +1,6021 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.18791694071220522,
+  "eval_steps": 500,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 1.0332,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 0.7996,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-06,
+      "loss": 0.9265,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 0.9685,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-06,
+      "loss": 0.9378,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6e-06,
+      "loss": 0.8732,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 0.7602,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 0.8929,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.4e-06,
+      "loss": 0.8032,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-06,
+      "loss": 0.7816,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.6e-06,
+      "loss": 0.9468,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.2e-06,
+      "loss": 1.0741,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.8e-06,
+      "loss": 0.8711,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 0.8376,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-06,
+      "loss": 0.8145,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 0.8266,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.02e-05,
+      "loss": 0.8253,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.08e-05,
+      "loss": 0.9463,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 0.9525,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-05,
+      "loss": 0.8613,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.26e-05,
+      "loss": 0.8314,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.32e-05,
+      "loss": 0.9321,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 0.8953,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.44e-05,
+      "loss": 0.7995,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-05,
+      "loss": 0.8905,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.56e-05,
+      "loss": 0.9119,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.62e-05,
+      "loss": 0.6862,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 0.9191,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.74e-05,
+      "loss": 0.877,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8e-05,
+      "loss": 0.7163,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.86e-05,
+      "loss": 0.7787,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 0.8967,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.98e-05,
+      "loss": 0.7819,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.04e-05,
+      "loss": 0.8425,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1e-05,
+      "loss": 0.7719,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.16e-05,
+      "loss": 0.9407,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.22e-05,
+      "loss": 0.8833,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 0.9032,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3400000000000003e-05,
+      "loss": 0.8153,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4e-05,
+      "loss": 0.7745,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4599999999999998e-05,
+      "loss": 0.8723,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.52e-05,
+      "loss": 0.7644,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.58e-05,
+      "loss": 0.8179,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.64e-05,
+      "loss": 0.8071,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 0.8132,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7600000000000003e-05,
+      "loss": 0.8378,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 0.8576,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.88e-05,
+      "loss": 0.8494,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.94e-05,
+      "loss": 0.5782,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-05,
+      "loss": 0.8991,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.06e-05,
+      "loss": 0.8166,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.12e-05,
+      "loss": 0.7971,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.18e-05,
+      "loss": 0.8448,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.24e-05,
+      "loss": 0.8292,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3e-05,
+      "loss": 0.8163,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3600000000000004e-05,
+      "loss": 0.8365,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.42e-05,
+      "loss": 0.7639,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.48e-05,
+      "loss": 0.8499,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.54e-05,
+      "loss": 0.8707,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6e-05,
+      "loss": 0.8381,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.66e-05,
+      "loss": 0.9216,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.72e-05,
+      "loss": 0.8795,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7800000000000004e-05,
+      "loss": 0.8209,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8400000000000005e-05,
+      "loss": 0.8344,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9e-05,
+      "loss": 0.8871,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.96e-05,
+      "loss": 0.728,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.02e-05,
+      "loss": 0.7665,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.08e-05,
+      "loss": 0.821,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.14e-05,
+      "loss": 0.7874,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2e-05,
+      "loss": 0.8541,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.26e-05,
+      "loss": 0.7519,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.32e-05,
+      "loss": 0.9324,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.38e-05,
+      "loss": 0.8668,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.44e-05,
+      "loss": 0.8193,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5e-05,
+      "loss": 0.8619,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5600000000000004e-05,
+      "loss": 0.8587,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6200000000000005e-05,
+      "loss": 0.7301,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6800000000000006e-05,
+      "loss": 0.7787,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.74e-05,
+      "loss": 0.8196,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8e-05,
+      "loss": 0.8312,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.86e-05,
+      "loss": 0.8714,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9199999999999997e-05,
+      "loss": 0.8062,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.98e-05,
+      "loss": 0.6693,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.04e-05,
+      "loss": 0.8631,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.1e-05,
+      "loss": 0.8563,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.16e-05,
+      "loss": 0.8991,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.22e-05,
+      "loss": 0.7307,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.28e-05,
+      "loss": 0.8957,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.3400000000000004e-05,
+      "loss": 0.8139,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4000000000000005e-05,
+      "loss": 0.9605,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4600000000000006e-05,
+      "loss": 0.8084,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.520000000000001e-05,
+      "loss": 0.8636,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.58e-05,
+      "loss": 0.7333,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6399999999999995e-05,
+      "loss": 0.854,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 0.8893,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.76e-05,
+      "loss": 0.9042,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.82e-05,
+      "loss": 0.8961,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.88e-05,
+      "loss": 0.8832,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.94e-05,
+      "loss": 0.8213,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6e-05,
+      "loss": 0.7678,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9999999670105504e-05,
+      "loss": 0.7245,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999999868042204e-05,
+      "loss": 0.8024,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9999997030949595e-05,
+      "loss": 0.8295,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9999994721688224e-05,
+      "loss": 0.7495,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999999175263799e-05,
+      "loss": 0.836,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999998812379895e-05,
+      "loss": 0.8252,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999998383517118e-05,
+      "loss": 0.7729,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9999978886754775e-05,
+      "loss": 0.7576,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999997327854984e-05,
+      "loss": 0.8935,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999996701055651e-05,
+      "loss": 0.8873,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999996008277492e-05,
+      "loss": 0.7484,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999995249520521e-05,
+      "loss": 0.757,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999994424784756e-05,
+      "loss": 0.8519,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999993534070214e-05,
+      "loss": 0.8135,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999992577376916e-05,
+      "loss": 0.7809,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999991554704882e-05,
+      "loss": 0.8468,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999990466054134e-05,
+      "loss": 0.9758,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999989311424698e-05,
+      "loss": 0.9353,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999988090816597e-05,
+      "loss": 0.8699,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9999868042298596e-05,
+      "loss": 0.874,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999985451664514e-05,
+      "loss": 0.6655,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999984033120589e-05,
+      "loss": 0.8626,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999982548598116e-05,
+      "loss": 0.7306,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999980998097128e-05,
+      "loss": 0.932,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999979381617659e-05,
+      "loss": 0.9458,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9999776991597445e-05,
+      "loss": 0.8097,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9999759507234204e-05,
+      "loss": 0.8559,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999974136308727e-05,
+      "loss": 0.8372,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999972255915705e-05,
+      "loss": 0.8184,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9999703095443926e-05,
+      "loss": 0.7572,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999968297194834e-05,
+      "loss": 0.9317,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9999662188670745e-05,
+      "loss": 0.9474,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999964074561159e-05,
+      "loss": 0.8571,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999961864277135e-05,
+      "loss": 0.8628,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.99995958801505e-05,
+      "loss": 0.734,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9999572457749545e-05,
+      "loss": 0.7531,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999954837556901e-05,
+      "loss": 0.7709,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999952363360941e-05,
+      "loss": 0.755,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9999498231871314e-05,
+      "loss": 0.8639,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999947217035525e-05,
+      "loss": 0.8272,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999944544906181e-05,
+      "loss": 0.8665,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999941806799158e-05,
+      "loss": 0.7484,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999939002714515e-05,
+      "loss": 0.9287,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999936132652315e-05,
+      "loss": 0.8519,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999933196612621e-05,
+      "loss": 0.93,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999930194595497e-05,
+      "loss": 0.6949,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999927126601009e-05,
+      "loss": 0.8102,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999923992629225e-05,
+      "loss": 0.9135,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9999207926802145e-05,
+      "loss": 0.8597,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9999175267540464e-05,
+      "loss": 0.7936,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9999141948507926e-05,
+      "loss": 0.9227,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999910796970527e-05,
+      "loss": 0.7899,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999907333113325e-05,
+      "loss": 0.8159,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999903803279262e-05,
+      "loss": 0.9484,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9999002074684153e-05,
+      "loss": 0.9407,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9998965456808646e-05,
+      "loss": 0.8982,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.99989281791669e-05,
+      "loss": 0.8212,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9998890241759736e-05,
+      "loss": 0.7662,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999885164458799e-05,
+      "loss": 0.8351,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9998812387652514e-05,
+      "loss": 0.7387,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999877247095416e-05,
+      "loss": 0.8321,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9998731894493826e-05,
+      "loss": 0.8062,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999869065827239e-05,
+      "loss": 0.705,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999864876229076e-05,
+      "loss": 0.8356,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999860620654985e-05,
+      "loss": 0.7994,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9998562991050606e-05,
+      "loss": 0.844,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999851911579398e-05,
+      "loss": 0.7925,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999847458078093e-05,
+      "loss": 0.7959,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999842938601244e-05,
+      "loss": 0.7132,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999838353148951e-05,
+      "loss": 0.7848,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999833701721314e-05,
+      "loss": 0.8137,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999828984318435e-05,
+      "loss": 0.8936,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999824200940418e-05,
+      "loss": 0.7018,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9998193515873685e-05,
+      "loss": 0.6827,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999814436259393e-05,
+      "loss": 0.9268,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9998094549566e-05,
+      "loss": 0.8198,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999804407679099e-05,
+      "loss": 0.8075,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999799294427e-05,
+      "loss": 0.8458,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9997941152004165e-05,
+      "loss": 0.8222,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999788869999462e-05,
+      "loss": 0.768,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9997835588242514e-05,
+      "loss": 0.7241,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999778181674903e-05,
+      "loss": 0.8091,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9997727385515334e-05,
+      "loss": 0.7761,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999767229454263e-05,
+      "loss": 0.7231,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999761654383214e-05,
+      "loss": 0.7994,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999756013338507e-05,
+      "loss": 0.8492,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999750306320267e-05,
+      "loss": 0.8025,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.99974453332862e-05,
+      "loss": 0.8548,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9997386943636924e-05,
+      "loss": 0.8746,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999732789425613e-05,
+      "loss": 0.9152,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.99972681851451e-05,
+      "loss": 0.8344,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9997207816305175e-05,
+      "loss": 0.8762,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999714678773767e-05,
+      "loss": 0.9612,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999708509944392e-05,
+      "loss": 0.7331,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9997022751425284e-05,
+      "loss": 0.7598,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999695974368315e-05,
+      "loss": 0.7899,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999689607621888e-05,
+      "loss": 0.7931,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999683174903389e-05,
+      "loss": 0.7744,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999676676212959e-05,
+      "loss": 0.8642,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.99967011155074e-05,
+      "loss": 0.7483,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999663480916878e-05,
+      "loss": 0.7995,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999656784311519e-05,
+      "loss": 0.7851,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999650021734808e-05,
+      "loss": 0.7556,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9996431931868954e-05,
+      "loss": 0.727,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9996362986679315e-05,
+      "loss": 0.8604,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9996293381780674e-05,
+      "loss": 0.7674,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999622311717456e-05,
+      "loss": 0.8444,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9996152192862526e-05,
+      "loss": 0.9209,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999608060884612e-05,
+      "loss": 0.8709,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9996008365126936e-05,
+      "loss": 0.6768,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999593546170654e-05,
+      "loss": 0.7567,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999586189858655e-05,
+      "loss": 0.8305,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999578767576858e-05,
+      "loss": 0.7292,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9995712793254255e-05,
+      "loss": 0.8674,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999563725104524e-05,
+      "loss": 0.8655,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999556104914317e-05,
+      "loss": 0.8689,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999548418754975e-05,
+      "loss": 0.8774,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999540666626665e-05,
+      "loss": 0.8134,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999532848529559e-05,
+      "loss": 0.8491,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999524964463827e-05,
+      "loss": 0.8354,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9995170144296436e-05,
+      "loss": 0.7627,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999508998427184e-05,
+      "loss": 1.0704,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999500916456624e-05,
+      "loss": 0.7801,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9994927685181415e-05,
+      "loss": 0.8051,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999484554611915e-05,
+      "loss": 0.8998,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999476274738126e-05,
+      "loss": 0.7718,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999467928896957e-05,
+      "loss": 0.8671,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.99945951708859e-05,
+      "loss": 0.8869,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999451039313211e-05,
+      "loss": 0.8355,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999442495571006e-05,
+      "loss": 0.9532,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9994338858621645e-05,
+      "loss": 0.768,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999425210186874e-05,
+      "loss": 0.8911,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999416468545326e-05,
+      "loss": 0.7113,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9994076609377115e-05,
+      "loss": 0.8187,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999398787364226e-05,
+      "loss": 0.9102,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999389847825065e-05,
+      "loss": 0.7694,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999380842320423e-05,
+      "loss": 0.7582,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9993717708505e-05,
+      "loss": 0.8563,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999362633415494e-05,
+      "loss": 0.8575,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.999353430015606e-05,
+      "loss": 0.7524,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.99934416065104e-05,
+      "loss": 0.7985,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.999334825321999e-05,
+      "loss": 0.7841,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.999325424028688e-05,
+      "loss": 0.8108,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.999315956771314e-05,
+      "loss": 0.866,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.999306423550085e-05,
+      "loss": 0.8494,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.999296824365211e-05,
+      "loss": 0.8344,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.999287159216903e-05,
+      "loss": 0.6927,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9992774281053726e-05,
+      "loss": 0.8304,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.999267631030836e-05,
+      "loss": 0.8693,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.999257767993507e-05,
+      "loss": 0.7813,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.999247838993603e-05,
+      "loss": 1.0163,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.999237844031342e-05,
+      "loss": 0.7866,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.999227783106945e-05,
+      "loss": 0.9145,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9992176562206313e-05,
+      "loss": 0.8601,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.999207463372625e-05,
+      "loss": 0.7739,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.99919720456315e-05,
+      "loss": 0.9324,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9991868797924325e-05,
+      "loss": 0.8729,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.999176489060699e-05,
+      "loss": 0.8107,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.999166032368177e-05,
+      "loss": 0.8985,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9991555097150986e-05,
+      "loss": 0.8163,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.999144921101693e-05,
+      "loss": 0.9045,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.999134266528195e-05,
+      "loss": 0.9768,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9991235459948375e-05,
+      "loss": 0.7006,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.999112759501857e-05,
+      "loss": 0.8501,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9991019070494914e-05,
+      "loss": 0.8564,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9990909886379776e-05,
+      "loss": 0.7695,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.999080004267557e-05,
+      "loss": 0.8575,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.999068953938471e-05,
+      "loss": 0.8781,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.999057837650962e-05,
+      "loss": 0.7221,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.999046655405276e-05,
+      "loss": 0.867,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.999035407201657e-05,
+      "loss": 0.8524,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.999024093040354e-05,
+      "loss": 0.7952,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.999012712921615e-05,
+      "loss": 0.8057,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.99900126684569e-05,
+      "loss": 0.85,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998989754812831e-05,
+      "loss": 0.8805,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998978176823292e-05,
+      "loss": 0.7014,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9989665328773274e-05,
+      "loss": 0.9181,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998954822975192e-05,
+      "loss": 0.7523,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9989430471171436e-05,
+      "loss": 0.8319,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998931205303444e-05,
+      "loss": 0.9637,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9989192975343487e-05,
+      "loss": 0.7763,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998907323810124e-05,
+      "loss": 0.8708,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998895284131031e-05,
+      "loss": 0.8101,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998883178497335e-05,
+      "loss": 0.8017,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9988710069093024e-05,
+      "loss": 0.8401,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998858769367201e-05,
+      "loss": 0.701,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9988464658713e-05,
+      "loss": 0.898,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998834096421869e-05,
+      "loss": 0.8737,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998821661019181e-05,
+      "loss": 0.8143,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9988091596635085e-05,
+      "loss": 0.7462,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998796592355128e-05,
+      "loss": 0.9357,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998783959094315e-05,
+      "loss": 0.8791,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9987712598813474e-05,
+      "loss": 0.8575,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.998758494716503e-05,
+      "loss": 0.8001,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.998745663600066e-05,
+      "loss": 0.8725,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9987327665323165e-05,
+      "loss": 0.8768,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.998719803513538e-05,
+      "loss": 0.8986,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9987067745440144e-05,
+      "loss": 0.8519,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.998693679624036e-05,
+      "loss": 0.8205,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9986805187538875e-05,
+      "loss": 0.8284,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.998667291933859e-05,
+      "loss": 0.8528,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9986539991642416e-05,
+      "loss": 0.7254,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.998640640445329e-05,
+      "loss": 0.7228,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9986272157774125e-05,
+      "loss": 0.9328,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.99861372516079e-05,
+      "loss": 0.8889,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.998600168595756e-05,
+      "loss": 0.8796,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9985865460826095e-05,
+      "loss": 0.8307,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.99857285762165e-05,
+      "loss": 0.7349,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.998559103213179e-05,
+      "loss": 0.8937,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.998545282857498e-05,
+      "loss": 0.7084,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9985313965549125e-05,
+      "loss": 0.9065,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9985174443057265e-05,
+      "loss": 0.7812,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.998503426110247e-05,
+      "loss": 0.8016,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9984893419687834e-05,
+      "loss": 0.7498,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.998475191881644e-05,
+      "loss": 0.8067,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.998460975849141e-05,
+      "loss": 0.8617,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.998446693871587e-05,
+      "loss": 0.7371,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9984323459492955e-05,
+      "loss": 0.7487,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9984179320825835e-05,
+      "loss": 0.8223,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9984034522717645e-05,
+      "loss": 0.9191,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.998388906517161e-05,
+      "loss": 0.8391,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9983742948190915e-05,
+      "loss": 0.8509,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9983596171778765e-05,
+      "loss": 0.7947,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.998344873593841e-05,
+      "loss": 0.766,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.998330064067306e-05,
+      "loss": 0.7974,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9983151885985985e-05,
+      "loss": 0.7885,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.998300247188047e-05,
+      "loss": 0.9425,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.998285239835979e-05,
+      "loss": 0.8962,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9982701665427254e-05,
+      "loss": 0.831,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.998255027308616e-05,
+      "loss": 0.8539,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.998239822133985e-05,
+      "loss": 0.7333,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9982245510191674e-05,
+      "loss": 0.7798,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.998209213964497e-05,
+      "loss": 0.8138,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9981938109703126e-05,
+      "loss": 0.797,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.998178342036953e-05,
+      "loss": 0.9049,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.998162807164758e-05,
+      "loss": 0.7876,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9981472063540697e-05,
+      "loss": 0.758,
+      "step": 337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9981315396052313e-05,
+      "loss": 0.776,
+      "step": 338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.998115806918586e-05,
+      "loss": 0.7658,
+      "step": 339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.998100008294481e-05,
+      "loss": 0.8095,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.998084143733264e-05,
+      "loss": 0.811,
+      "step": 341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.998068213235282e-05,
+      "loss": 0.883,
+      "step": 342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.998052216800888e-05,
+      "loss": 0.8393,
+      "step": 343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9980361544304327e-05,
+      "loss": 0.8455,
+      "step": 344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9980200261242686e-05,
+      "loss": 0.7321,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.998003831882751e-05,
+      "loss": 0.826,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997987571706237e-05,
+      "loss": 0.729,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9979712455950826e-05,
+      "loss": 0.8787,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9979548535496474e-05,
+      "loss": 0.7997,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997938395570293e-05,
+      "loss": 0.7883,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.99792187165738e-05,
+      "loss": 0.8215,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997905281811272e-05,
+      "loss": 0.8128,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997888626032334e-05,
+      "loss": 0.7826,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997871904320933e-05,
+      "loss": 0.8116,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997855116677436e-05,
+      "loss": 0.7321,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9978382631022116e-05,
+      "loss": 0.9785,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997821343595632e-05,
+      "loss": 0.7587,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997804358158068e-05,
+      "loss": 0.7869,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997787306789895e-05,
+      "loss": 0.9027,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9977701894914855e-05,
+      "loss": 0.7536,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997753006263218e-05,
+      "loss": 0.8296,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997735757105469e-05,
+      "loss": 0.7669,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9977184420186186e-05,
+      "loss": 0.7666,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997701061003048e-05,
+      "loss": 0.7532,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997683614059138e-05,
+      "loss": 0.8618,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9976661011872743e-05,
+      "loss": 0.7741,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.99764852238784e-05,
+      "loss": 0.7539,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997630877661224e-05,
+      "loss": 0.7913,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9976131670078117e-05,
+      "loss": 0.793,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997595390427995e-05,
+      "loss": 0.7956,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997577547922164e-05,
+      "loss": 0.8038,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.99755963949071e-05,
+      "loss": 0.8854,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997541665134028e-05,
+      "loss": 0.7581,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9975236248525125e-05,
+      "loss": 0.8757,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997505518646562e-05,
+      "loss": 0.7842,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9974873465165734e-05,
+      "loss": 0.8536,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997469108462947e-05,
+      "loss": 0.8853,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9974508044860825e-05,
+      "loss": 0.7615,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997432434586384e-05,
+      "loss": 0.8048,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997413998764255e-05,
+      "loss": 0.7203,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997395497020101e-05,
+      "loss": 0.8504,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997376929354328e-05,
+      "loss": 0.8546,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997358295767346e-05,
+      "loss": 0.8185,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997339596259563e-05,
+      "loss": 0.7988,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997320830831392e-05,
+      "loss": 0.7517,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9973019994832446e-05,
+      "loss": 0.837,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997283102215536e-05,
+      "loss": 0.8774,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.99726413902868e-05,
+      "loss": 0.78,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997245109923096e-05,
+      "loss": 0.9168,
+      "step": 389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9972260148992e-05,
+      "loss": 0.8345,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9972068539574143e-05,
+      "loss": 0.834,
+      "step": 391
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997187627098158e-05,
+      "loss": 0.7731,
+      "step": 392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9971683343218564e-05,
+      "loss": 0.8661,
+      "step": 393
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997148975628932e-05,
+      "loss": 0.7246,
+      "step": 394
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997129551019811e-05,
+      "loss": 0.8897,
+      "step": 395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997110060494921e-05,
+      "loss": 0.8914,
+      "step": 396
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.99709050405469e-05,
+      "loss": 0.7198,
+      "step": 397
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997070881699549e-05,
+      "loss": 0.7748,
+      "step": 398
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.997051193429929e-05,
+      "loss": 0.7132,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.997031439246263e-05,
+      "loss": 0.7825,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.997011619148985e-05,
+      "loss": 0.785,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.996991733138532e-05,
+      "loss": 0.7937,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.99697178121534e-05,
+      "loss": 0.785,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.996951763379849e-05,
+      "loss": 0.7723,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9969316796325e-05,
+      "loss": 0.7915,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9969115299737315e-05,
+      "loss": 0.6737,
+      "step": 406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.99689131440399e-05,
+      "loss": 0.919,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.996871032923718e-05,
+      "loss": 0.792,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.996850685533363e-05,
+      "loss": 0.8157,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.996830272233371e-05,
+      "loss": 0.8379,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.996809793024193e-05,
+      "loss": 0.8289,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.996789247906277e-05,
+      "loss": 0.717,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.996768636880077e-05,
+      "loss": 0.7738,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.996747959946045e-05,
+      "loss": 0.797,
+      "step": 414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.996727217104635e-05,
+      "loss": 0.792,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.996706408356305e-05,
+      "loss": 0.8767,
+      "step": 416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9966855337015125e-05,
+      "loss": 0.8576,
+      "step": 417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.996664593140715e-05,
+      "loss": 0.8336,
+      "step": 418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9966435866743754e-05,
+      "loss": 0.7237,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.996622514302953e-05,
+      "loss": 0.7795,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.996601376026914e-05,
+      "loss": 0.8923,
+      "step": 421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.996580171846721e-05,
+      "loss": 0.7606,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9965589017628406e-05,
+      "loss": 0.6944,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.996537565775742e-05,
+      "loss": 0.9031,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9965161638858933e-05,
+      "loss": 0.7613,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9964946960937654e-05,
+      "loss": 0.6995,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9964731623998313e-05,
+      "loss": 0.9074,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9964515628045636e-05,
+      "loss": 0.7229,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.996429897308438e-05,
+      "loss": 0.807,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.996408165911929e-05,
+      "loss": 0.7672,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9963863686155176e-05,
+      "loss": 0.8736,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9963645054196806e-05,
+      "loss": 0.8468,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.996342576324901e-05,
+      "loss": 0.8667,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.99632058133166e-05,
+      "loss": 0.8528,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.996298520440441e-05,
+      "loss": 0.801,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.996276393651729e-05,
+      "loss": 0.7381,
+      "step": 436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.996254200966012e-05,
+      "loss": 0.8424,
+      "step": 437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9962319423837776e-05,
+      "loss": 0.7281,
+      "step": 438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9962096179055144e-05,
+      "loss": 0.7163,
+      "step": 439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.996187227531714e-05,
+      "loss": 0.8709,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9961647712628694e-05,
+      "loss": 0.6349,
+      "step": 441
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9961422490994735e-05,
+      "loss": 0.7883,
+      "step": 442
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9961196610420216e-05,
+      "loss": 0.7439,
+      "step": 443
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.996097007091011e-05,
+      "loss": 0.7658,
+      "step": 444
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.996074287246941e-05,
+      "loss": 0.8353,
+      "step": 445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.996051501510309e-05,
+      "loss": 0.7777,
+      "step": 446
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9960286498816175e-05,
+      "loss": 0.717,
+      "step": 447
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9960057323613685e-05,
+      "loss": 0.888,
+      "step": 448
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9959827489500674e-05,
+      "loss": 0.7494,
+      "step": 449
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9959596996482184e-05,
+      "loss": 0.8449,
+      "step": 450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9959365844563275e-05,
+      "loss": 0.8346,
+      "step": 451
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9959134033749045e-05,
+      "loss": 0.6486,
+      "step": 452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.99589015640446e-05,
+      "loss": 0.7517,
+      "step": 453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9958668435455035e-05,
+      "loss": 0.732,
+      "step": 454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.995843464798548e-05,
+      "loss": 0.9455,
+      "step": 455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9958200201641084e-05,
+      "loss": 0.8166,
+      "step": 456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9957965096427e-05,
+      "loss": 0.7384,
+      "step": 457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.995772933234841e-05,
+      "loss": 0.7982,
+      "step": 458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9957492909410474e-05,
+      "loss": 0.7755,
+      "step": 459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.995725582761841e-05,
+      "loss": 0.846,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.995701808697742e-05,
+      "loss": 0.6818,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.995677968749275e-05,
+      "loss": 0.7076,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.995654062916964e-05,
+      "loss": 0.8118,
+      "step": 463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9956300912013324e-05,
+      "loss": 0.8096,
+      "step": 464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9956060536029104e-05,
+      "loss": 0.8259,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9955819501222244e-05,
+      "loss": 0.705,
+      "step": 466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9955577807598064e-05,
+      "loss": 0.8215,
+      "step": 467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.995533545516186e-05,
+      "loss": 0.7097,
+      "step": 468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.995509244391898e-05,
+      "loss": 0.6895,
+      "step": 469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.995484877387476e-05,
+      "loss": 0.6963,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.995460444503456e-05,
+      "loss": 0.8469,
+      "step": 471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.995435945740376e-05,
+      "loss": 0.7172,
+      "step": 472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.995411381098772e-05,
+      "loss": 1.031,
+      "step": 473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.995386750579188e-05,
+      "loss": 0.781,
+      "step": 474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.995362054182164e-05,
+      "loss": 0.8177,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.995337291908243e-05,
+      "loss": 0.8519,
+      "step": 476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9953124637579696e-05,
+      "loss": 0.6667,
+      "step": 477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.99528756973189e-05,
+      "loss": 0.8701,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.995262609830553e-05,
+      "loss": 0.8074,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.995237584054505e-05,
+      "loss": 0.8239,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9952124924042984e-05,
+      "loss": 0.8743,
+      "step": 481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.995187334880485e-05,
+      "loss": 0.6037,
+      "step": 482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.995162111483616e-05,
+      "loss": 0.8036,
+      "step": 483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9951368222142484e-05,
+      "loss": 0.8537,
+      "step": 484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.995111467072937e-05,
+      "loss": 0.8805,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.99508604606024e-05,
+      "loss": 0.7512,
+      "step": 486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.995060559176717e-05,
+      "loss": 0.7919,
+      "step": 487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.995035006422928e-05,
+      "loss": 0.8408,
+      "step": 488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.995009387799435e-05,
+      "loss": 0.7472,
+      "step": 489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9949837033068005e-05,
+      "loss": 0.7319,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.994957952945591e-05,
+      "loss": 0.8333,
+      "step": 491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.994932136716372e-05,
+      "loss": 0.8813,
+      "step": 492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9949062546197115e-05,
+      "loss": 0.9251,
+      "step": 493
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.994880306656179e-05,
+      "loss": 0.732,
+      "step": 494
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9948542928263446e-05,
+      "loss": 0.8163,
+      "step": 495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.99482821313078e-05,
+      "loss": 0.7552,
+      "step": 496
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.99480206757006e-05,
+      "loss": 0.7811,
+      "step": 497
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.994775856144759e-05,
+      "loss": 0.8851,
+      "step": 498
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.994749578855453e-05,
+      "loss": 0.7862,
+      "step": 499
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.994723235702721e-05,
+      "loss": 0.8771,
+      "step": 500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.994696826687141e-05,
+      "loss": 0.8973,
+      "step": 501
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.994670351809295e-05,
+      "loss": 0.8543,
+      "step": 502
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.994643811069765e-05,
+      "loss": 0.8006,
+      "step": 503
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.994617204469134e-05,
+      "loss": 0.866,
+      "step": 504
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9945905320079885e-05,
+      "loss": 0.7474,
+      "step": 505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.994563793686914e-05,
+      "loss": 0.804,
+      "step": 506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9945369895064986e-05,
+      "loss": 0.8908,
+      "step": 507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9945101194673324e-05,
+      "loss": 0.8367,
+      "step": 508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.994483183570005e-05,
+      "loss": 0.8643,
+      "step": 509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.994456181815111e-05,
+      "loss": 0.7302,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9944291142032426e-05,
+      "loss": 0.879,
+      "step": 511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9944019807349954e-05,
+      "loss": 0.8082,
+      "step": 512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9943747814109666e-05,
+      "loss": 0.7482,
+      "step": 513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.994347516231754e-05,
+      "loss": 0.8469,
+      "step": 514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9943201851979574e-05,
+      "loss": 0.8149,
+      "step": 515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.994292788310178e-05,
+      "loss": 0.8049,
+      "step": 516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.994265325569018e-05,
+      "loss": 0.9191,
+      "step": 517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.994237796975081e-05,
+      "loss": 0.8181,
+      "step": 518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.994210202528973e-05,
+      "loss": 0.7609,
+      "step": 519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.994182542231302e-05,
+      "loss": 0.7077,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.994154816082674e-05,
+      "loss": 0.8684,
+      "step": 521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9941270240837006e-05,
+      "loss": 0.841,
+      "step": 522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9940991662349925e-05,
+      "loss": 0.7711,
+      "step": 523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9940712425371625e-05,
+      "loss": 0.769,
+      "step": 524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.994043252990824e-05,
+      "loss": 0.8333,
+      "step": 525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.994015197596593e-05,
+      "loss": 0.8188,
+      "step": 526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.993987076355087e-05,
+      "loss": 0.7714,
+      "step": 527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.993958889266924e-05,
+      "loss": 0.874,
+      "step": 528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9939306363327234e-05,
+      "loss": 0.7457,
+      "step": 529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.993902317553108e-05,
+      "loss": 0.7106,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.993873932928699e-05,
+      "loss": 0.741,
+      "step": 531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9938454824601214e-05,
+      "loss": 0.7118,
+      "step": 532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.993816966148002e-05,
+      "loss": 0.7757,
+      "step": 533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.993788383992966e-05,
+      "loss": 0.8359,
+      "step": 534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9937597359956435e-05,
+      "loss": 0.7857,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9937310221566635e-05,
+      "loss": 0.742,
+      "step": 536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9937022424766585e-05,
+      "loss": 0.7601,
+      "step": 537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.99367339695626e-05,
+      "loss": 0.78,
+      "step": 538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9936444855961045e-05,
+      "loss": 0.7159,
+      "step": 539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.993615508396826e-05,
+      "loss": 0.7284,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9935864653590624e-05,
+      "loss": 0.8424,
+      "step": 541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.993557356483453e-05,
+      "loss": 0.8108,
+      "step": 542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.993528181770637e-05,
+      "loss": 1.0558,
+      "step": 543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9934989412212566e-05,
+      "loss": 0.9712,
+      "step": 544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.993469634835955e-05,
+      "loss": 0.9326,
+      "step": 545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.993440262615376e-05,
+      "loss": 0.7347,
+      "step": 546
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.993410824560167e-05,
+      "loss": 0.8127,
+      "step": 547
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.993381320670974e-05,
+      "loss": 0.8716,
+      "step": 548
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.993351750948447e-05,
+      "loss": 0.9098,
+      "step": 549
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.993322115393235e-05,
+      "loss": 0.8756,
+      "step": 550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9932924140059916e-05,
+      "loss": 0.7015,
+      "step": 551
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.993262646787368e-05,
+      "loss": 0.8441,
+      "step": 552
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.99323281373802e-05,
+      "loss": 0.9761,
+      "step": 553
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.993202914858603e-05,
+      "loss": 0.7554,
+      "step": 554
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9931729501497756e-05,
+      "loss": 0.8423,
+      "step": 555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.993142919612197e-05,
+      "loss": 0.8332,
+      "step": 556
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.993112823246526e-05,
+      "loss": 0.7036,
+      "step": 557
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9930826610534264e-05,
+      "loss": 0.8263,
+      "step": 558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9930524330335596e-05,
+      "loss": 0.8945,
+      "step": 559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9930221391875926e-05,
+      "loss": 0.7429,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.99299177951619e-05,
+      "loss": 0.747,
+      "step": 561
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.99296135402002e-05,
+      "loss": 0.8044,
+      "step": 562
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.992930862699752e-05,
+      "loss": 0.834,
+      "step": 563
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.992900305556056e-05,
+      "loss": 0.7906,
+      "step": 564
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.992869682589605e-05,
+      "loss": 0.9249,
+      "step": 565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9928389938010707e-05,
+      "loss": 0.7057,
+      "step": 566
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.99280823919113e-05,
+      "loss": 0.8805,
+      "step": 567
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9927774187604585e-05,
+      "loss": 0.887,
+      "step": 568
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.992746532509734e-05,
+      "loss": 0.8601,
+      "step": 569
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.992715580439636e-05,
+      "loss": 0.8494,
+      "step": 570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.992684562550844e-05,
+      "loss": 0.959,
+      "step": 571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.992653478844042e-05,
+      "loss": 0.8656,
+      "step": 572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.992622329319913e-05,
+      "loss": 0.8579,
+      "step": 573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9925911139791416e-05,
+      "loss": 0.8319,
+      "step": 574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.992559832822416e-05,
+      "loss": 0.7723,
+      "step": 575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.992528485850421e-05,
+      "loss": 0.8531,
+      "step": 576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.992497073063848e-05,
+      "loss": 0.9316,
+      "step": 577
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9924655944633887e-05,
+      "loss": 0.8842,
+      "step": 578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9924340500497336e-05,
+      "loss": 0.8072,
+      "step": 579
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9924024398235776e-05,
+      "loss": 0.831,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.992370763785616e-05,
+      "loss": 0.9104,
+      "step": 581
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9923390219365435e-05,
+      "loss": 0.8886,
+      "step": 582
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.99230721427706e-05,
+      "loss": 0.6671,
+      "step": 583
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9922753408078654e-05,
+      "loss": 0.7643,
+      "step": 584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.99224340152966e-05,
+      "loss": 0.7602,
+      "step": 585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.992211396443146e-05,
+      "loss": 0.7737,
+      "step": 586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.992179325549028e-05,
+      "loss": 0.8805,
+      "step": 587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9921471888480105e-05,
+      "loss": 0.7505,
+      "step": 588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9921149863408004e-05,
+      "loss": 0.7997,
+      "step": 589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.992082718028107e-05,
+      "loss": 0.8676,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9920503839106384e-05,
+      "loss": 0.8555,
+      "step": 591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.992017983989107e-05,
+      "loss": 0.7605,
+      "step": 592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9919855182642244e-05,
+      "loss": 0.8968,
+      "step": 593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.991952986736705e-05,
+      "loss": 0.8512,
+      "step": 594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9919203894072646e-05,
+      "loss": 0.7587,
+      "step": 595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.99188772627662e-05,
+      "loss": 0.909,
+      "step": 596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.99185499734549e-05,
+      "loss": 0.8389,
+      "step": 597
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9918222026145934e-05,
+      "loss": 0.8794,
+      "step": 598
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.991789342084651e-05,
+      "loss": 0.8615,
+      "step": 599
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.991756415756387e-05,
+      "loss": 0.7219,
+      "step": 600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.991723423630525e-05,
+      "loss": 0.8113,
+      "step": 601
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9916903657077905e-05,
+      "loss": 0.8344,
+      "step": 602
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.99165724198891e-05,
+      "loss": 0.6734,
+      "step": 603
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.991624052474613e-05,
+      "loss": 0.7266,
+      "step": 604
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9915907971656294e-05,
+      "loss": 0.7912,
+      "step": 605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9915574760626894e-05,
+      "loss": 1.0072,
+      "step": 606
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.991524089166527e-05,
+      "loss": 0.9409,
+      "step": 607
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.991490636477876e-05,
+      "loss": 0.8279,
+      "step": 608
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9914571179974726e-05,
+      "loss": 0.7029,
+      "step": 609
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.991423533726053e-05,
+      "loss": 0.8824,
+      "step": 610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.991389883664357e-05,
+      "loss": 0.8082,
+      "step": 611
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.991356167813124e-05,
+      "loss": 0.7994,
+      "step": 612
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9913223861730945e-05,
+      "loss": 0.7128,
+      "step": 613
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9912885387450135e-05,
+      "loss": 0.6518,
+      "step": 614
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9912546255296244e-05,
+      "loss": 0.8102,
+      "step": 615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9912206465276724e-05,
+      "loss": 0.8845,
+      "step": 616
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9911866017399065e-05,
+      "loss": 0.8356,
+      "step": 617
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.991152491167073e-05,
+      "loss": 0.8054,
+      "step": 618
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.991118314809925e-05,
+      "loss": 0.8733,
+      "step": 619
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9910840726692115e-05,
+      "loss": 0.7696,
+      "step": 620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9910497647456874e-05,
+      "loss": 0.7626,
+      "step": 621
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.991015391040106e-05,
+      "loss": 0.7455,
+      "step": 622
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.990980951553224e-05,
+      "loss": 0.7952,
+      "step": 623
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.990946446285799e-05,
+      "loss": 0.786,
+      "step": 624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.99091187523859e-05,
+      "loss": 0.7848,
+      "step": 625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.990877238412356e-05,
+      "loss": 0.7795,
+      "step": 626
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9908425358078603e-05,
+      "loss": 0.7499,
+      "step": 627
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.990807767425865e-05,
+      "loss": 0.9152,
+      "step": 628
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.990772933267135e-05,
+      "loss": 0.8096,
+      "step": 629
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.990738033332437e-05,
+      "loss": 0.8836,
+      "step": 630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.990703067622539e-05,
+      "loss": 0.783,
+      "step": 631
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9906680361382086e-05,
+      "loss": 0.689,
+      "step": 632
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.990632938880217e-05,
+      "loss": 0.7406,
+      "step": 633
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.990597775849335e-05,
+      "loss": 0.7689,
+      "step": 634
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9905625470463376e-05,
+      "loss": 0.656,
+      "step": 635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9905272524719984e-05,
+      "loss": 0.7853,
+      "step": 636
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.990491892127094e-05,
+      "loss": 0.8801,
+      "step": 637
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.990456466012402e-05,
+      "loss": 0.7618,
+      "step": 638
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.990420974128702e-05,
+      "loss": 0.7529,
+      "step": 639
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.990385416476775e-05,
+      "loss": 0.6657,
+      "step": 640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.990349793057401e-05,
+      "loss": 0.8902,
+      "step": 641
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.990314103871365e-05,
+      "loss": 0.8582,
+      "step": 642
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.990278348919452e-05,
+      "loss": 0.8811,
+      "step": 643
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9902425282024484e-05,
+      "loss": 0.8655,
+      "step": 644
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.990206641721141e-05,
+      "loss": 0.7528,
+      "step": 645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.99017068947632e-05,
+      "loss": 0.8776,
+      "step": 646
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.990134671468775e-05,
+      "loss": 0.8457,
+      "step": 647
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9900985876993e-05,
+      "loss": 0.9195,
+      "step": 648
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.990062438168687e-05,
+      "loss": 0.8502,
+      "step": 649
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9900262228777306e-05,
+      "loss": 0.7403,
+      "step": 650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.989989941827229e-05,
+      "loss": 0.722,
+      "step": 651
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9899535950179795e-05,
+      "loss": 0.8618,
+      "step": 652
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.989917182450781e-05,
+      "loss": 0.8072,
+      "step": 653
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.989880704126435e-05,
+      "loss": 0.894,
+      "step": 654
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.989844160045743e-05,
+      "loss": 0.8953,
+      "step": 655
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.98980755020951e-05,
+      "loss": 0.9292,
+      "step": 656
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9897708746185383e-05,
+      "loss": 0.6746,
+      "step": 657
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.989734133273638e-05,
+      "loss": 0.8496,
+      "step": 658
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.989697326175615e-05,
+      "loss": 0.8861,
+      "step": 659
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.989660453325281e-05,
+      "loss": 0.7242,
+      "step": 660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.989623514723443e-05,
+      "loss": 0.8102,
+      "step": 661
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.989586510370917e-05,
+      "loss": 0.8024,
+      "step": 662
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.989549440268515e-05,
+      "loss": 0.767,
+      "step": 663
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.989512304417053e-05,
+      "loss": 0.687,
+      "step": 664
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9894751028173475e-05,
+      "loss": 0.7871,
+      "step": 665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9894378354702175e-05,
+      "loss": 0.7369,
+      "step": 666
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.989400502376482e-05,
+      "loss": 0.727,
+      "step": 667
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9893631035369606e-05,
+      "loss": 0.7842,
+      "step": 668
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9893256389524775e-05,
+      "loss": 0.8541,
+      "step": 669
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.989288108623858e-05,
+      "loss": 0.9089,
+      "step": 670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.989250512551924e-05,
+      "loss": 0.7534,
+      "step": 671
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9892128507375045e-05,
+      "loss": 0.5936,
+      "step": 672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9891751231814284e-05,
+      "loss": 0.6942,
+      "step": 673
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.989137329884523e-05,
+      "loss": 0.8069,
+      "step": 674
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.989099470847622e-05,
+      "loss": 0.736,
+      "step": 675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9890615460715565e-05,
+      "loss": 0.7954,
+      "step": 676
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.989023555557162e-05,
+      "loss": 0.8147,
+      "step": 677
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9889854993052725e-05,
+      "loss": 0.8035,
+      "step": 678
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.988947377316725e-05,
+      "loss": 0.7567,
+      "step": 679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.98890918959236e-05,
+      "loss": 0.8222,
+      "step": 680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.988870936133014e-05,
+      "loss": 0.7346,
+      "step": 681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.988832616939532e-05,
+      "loss": 0.7342,
+      "step": 682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.988794232012754e-05,
+      "loss": 0.8605,
+      "step": 683
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.988755781353525e-05,
+      "loss": 0.8759,
+      "step": 684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9887172649626906e-05,
+      "loss": 0.8362,
+      "step": 685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.988678682841099e-05,
+      "loss": 0.8329,
+      "step": 686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.988640034989597e-05,
+      "loss": 0.8374,
+      "step": 687
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9886013214090353e-05,
+      "loss": 0.7762,
+      "step": 688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.988562542100266e-05,
+      "loss": 0.7888,
+      "step": 689
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.988523697064141e-05,
+      "loss": 0.7795,
+      "step": 690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.988484786301516e-05,
+      "loss": 0.7457,
+      "step": 691
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9884458098132444e-05,
+      "loss": 0.7969,
+      "step": 692
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.988406767600186e-05,
+      "loss": 0.8538,
+      "step": 693
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.988367659663197e-05,
+      "loss": 0.7924,
+      "step": 694
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.988328486003139e-05,
+      "loss": 0.7931,
+      "step": 695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.988289246620874e-05,
+      "loss": 0.8222,
+      "step": 696
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.988249941517263e-05,
+      "loss": 0.6647,
+      "step": 697
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.988210570693173e-05,
+      "loss": 0.7299,
+      "step": 698
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.988171134149468e-05,
+      "loss": 0.7012,
+      "step": 699
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.988131631887016e-05,
+      "loss": 0.7706,
+      "step": 700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.988092063906685e-05,
+      "loss": 0.9783,
+      "step": 701
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.988052430209346e-05,
+      "loss": 0.7894,
+      "step": 702
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.988012730795871e-05,
+      "loss": 0.7916,
+      "step": 703
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.987972965667133e-05,
+      "loss": 0.802,
+      "step": 704
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.987933134824005e-05,
+      "loss": 0.7928,
+      "step": 705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.987893238267366e-05,
+      "loss": 0.7272,
+      "step": 706
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9878532759980905e-05,
+      "loss": 0.8978,
+      "step": 707
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.987813248017059e-05,
+      "loss": 0.8089,
+      "step": 708
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.987773154325151e-05,
+      "loss": 0.8662,
+      "step": 709
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.987732994923249e-05,
+      "loss": 0.7808,
+      "step": 710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9876927698122356e-05,
+      "loss": 0.7293,
+      "step": 711
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.987652478992995e-05,
+      "loss": 0.7769,
+      "step": 712
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9876121224664164e-05,
+      "loss": 0.7572,
+      "step": 713
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9875717002333826e-05,
+      "loss": 0.7736,
+      "step": 714
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.987531212294786e-05,
+      "loss": 0.7953,
+      "step": 715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.987490658651516e-05,
+      "loss": 0.787,
+      "step": 716
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.987450039304465e-05,
+      "loss": 0.8052,
+      "step": 717
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.987409354254526e-05,
+      "loss": 0.6961,
+      "step": 718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.987368603502593e-05,
+      "loss": 0.8254,
+      "step": 719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.987327787049563e-05,
+      "loss": 0.8433,
+      "step": 720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.987286904896333e-05,
+      "loss": 0.7642,
+      "step": 721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.987245957043804e-05,
+      "loss": 0.7958,
+      "step": 722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9872049434928744e-05,
+      "loss": 0.9232,
+      "step": 723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.987163864244448e-05,
+      "loss": 0.8804,
+      "step": 724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9871227192994266e-05,
+      "loss": 0.7168,
+      "step": 725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.987081508658716e-05,
+      "loss": 0.8434,
+      "step": 726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9870402323232225e-05,
+      "loss": 0.8671,
+      "step": 727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.986998890293854e-05,
+      "loss": 0.8608,
+      "step": 728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.986957482571519e-05,
+      "loss": 0.8753,
+      "step": 729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.986916009157129e-05,
+      "loss": 0.8721,
+      "step": 730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9868744700515965e-05,
+      "loss": 0.7826,
+      "step": 731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.986832865255833e-05,
+      "loss": 0.936,
+      "step": 732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9867911947707556e-05,
+      "loss": 0.81,
+      "step": 733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.98674945859728e-05,
+      "loss": 0.8034,
+      "step": 734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.986707656736324e-05,
+      "loss": 0.8335,
+      "step": 735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9866657891888075e-05,
+      "loss": 0.8413,
+      "step": 736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9866238559556514e-05,
+      "loss": 0.8161,
+      "step": 737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.986581857037776e-05,
+      "loss": 0.7233,
+      "step": 738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9865397924361075e-05,
+      "loss": 0.7087,
+      "step": 739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9864976621515695e-05,
+      "loss": 0.6731,
+      "step": 740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9864554661850895e-05,
+      "loss": 0.7319,
+      "step": 741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9864132045375944e-05,
+      "loss": 0.8738,
+      "step": 742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.986370877210014e-05,
+      "loss": 0.7161,
+      "step": 743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9863284842032806e-05,
+      "loss": 0.8402,
+      "step": 744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.986286025518325e-05,
+      "loss": 0.9499,
+      "step": 745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.986243501156081e-05,
+      "loss": 0.7786,
+      "step": 746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9862009111174845e-05,
+      "loss": 0.6824,
+      "step": 747
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9861582554034724e-05,
+      "loss": 0.7925,
+      "step": 748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.986115534014982e-05,
+      "loss": 0.7308,
+      "step": 749
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.986072746952953e-05,
+      "loss": 0.675,
+      "step": 750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.986029894218326e-05,
+      "loss": 0.7757,
+      "step": 751
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9859869758120456e-05,
+      "loss": 0.94,
+      "step": 752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.985943991735054e-05,
+      "loss": 0.8009,
+      "step": 753
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.985900941988296e-05,
+      "loss": 0.8321,
+      "step": 754
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.98585782657272e-05,
+      "loss": 0.9798,
+      "step": 755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.985814645489272e-05,
+      "loss": 0.8476,
+      "step": 756
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.985771398738905e-05,
+      "loss": 0.8846,
+      "step": 757
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.985728086322567e-05,
+      "loss": 0.8774,
+      "step": 758
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.985684708241212e-05,
+      "loss": 0.7096,
+      "step": 759
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.985641264495794e-05,
+      "loss": 0.8164,
+      "step": 760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9855977550872676e-05,
+      "loss": 0.7679,
+      "step": 761
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.985554180016591e-05,
+      "loss": 0.8591,
+      "step": 762
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9855105392847216e-05,
+      "loss": 0.8316,
+      "step": 763
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.985466832892619e-05,
+      "loss": 0.9169,
+      "step": 764
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.985423060841246e-05,
+      "loss": 0.7841,
+      "step": 765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9853792231315636e-05,
+      "loss": 0.8097,
+      "step": 766
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9853353197645363e-05,
+      "loss": 0.7585,
+      "step": 767
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.98529135074113e-05,
+      "loss": 0.8807,
+      "step": 768
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.985247316062311e-05,
+      "loss": 0.7894,
+      "step": 769
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9852032157290505e-05,
+      "loss": 0.9108,
+      "step": 770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.985159049742314e-05,
+      "loss": 0.8792,
+      "step": 771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.985114818103076e-05,
+      "loss": 0.7926,
+      "step": 772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9850705208123076e-05,
+      "loss": 0.7996,
+      "step": 773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.985026157870984e-05,
+      "loss": 0.8963,
+      "step": 774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9849817292800815e-05,
+      "loss": 0.8074,
+      "step": 775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.984937235040576e-05,
+      "loss": 0.8583,
+      "step": 776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.984892675153447e-05,
+      "loss": 0.7128,
+      "step": 777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9848480496196725e-05,
+      "loss": 0.769,
+      "step": 778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9848033584402356e-05,
+      "loss": 0.7741,
+      "step": 779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.984758601616119e-05,
+      "loss": 0.7902,
+      "step": 780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.984713779148308e-05,
+      "loss": 0.7969,
+      "step": 781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.984668891037787e-05,
+      "loss": 0.8908,
+      "step": 782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9846239372855436e-05,
+      "loss": 0.834,
+      "step": 783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.984578917892566e-05,
+      "loss": 0.8202,
+      "step": 784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9845338328598455e-05,
+      "loss": 0.8281,
+      "step": 785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9844886821883735e-05,
+      "loss": 0.7191,
+      "step": 786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.984443465879141e-05,
+      "loss": 0.7581,
+      "step": 787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9843981839331446e-05,
+      "loss": 0.8448,
+      "step": 788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.984352836351379e-05,
+      "loss": 0.7852,
+      "step": 789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.984307423134843e-05,
+      "loss": 0.8379,
+      "step": 790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.984261944284533e-05,
+      "loss": 0.739,
+      "step": 791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.984216399801452e-05,
+      "loss": 0.8135,
+      "step": 792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.984170789686599e-05,
+      "loss": 0.8544,
+      "step": 793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.984125113940979e-05,
+      "loss": 0.7238,
+      "step": 794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.984079372565595e-05,
+      "loss": 0.8356,
+      "step": 795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9840335655614546e-05,
+      "loss": 0.833,
+      "step": 796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9839876929295646e-05,
+      "loss": 0.8089,
+      "step": 797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.983941754670934e-05,
+      "loss": 0.7677,
+      "step": 798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.983895750786572e-05,
+      "loss": 0.7308,
+      "step": 799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9838496812774926e-05,
+      "loss": 0.8211,
+      "step": 800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9838035461447064e-05,
+      "loss": 0.8127,
+      "step": 801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9837573453892306e-05,
+      "loss": 0.7816,
+      "step": 802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.983711079012078e-05,
+      "loss": 0.8093,
+      "step": 803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.98366474701427e-05,
+      "loss": 0.8776,
+      "step": 804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.983618349396823e-05,
+      "loss": 0.6925,
+      "step": 805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.983571886160759e-05,
+      "loss": 0.8239,
+      "step": 806
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.983525357307098e-05,
+      "loss": 0.7842,
+      "step": 807
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.983478762836864e-05,
+      "loss": 0.6514,
+      "step": 808
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9834321027510825e-05,
+      "loss": 0.7474,
+      "step": 809
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.98338537705078e-05,
+      "loss": 0.7652,
+      "step": 810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.983338585736983e-05,
+      "loss": 0.6694,
+      "step": 811
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.98329172881072e-05,
+      "loss": 0.8169,
+      "step": 812
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9832448062730235e-05,
+      "loss": 0.7094,
+      "step": 813
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.983197818124924e-05,
+      "loss": 0.7934,
+      "step": 814
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.983150764367456e-05,
+      "loss": 0.7287,
+      "step": 815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9831036450016525e-05,
+      "loss": 0.7709,
+      "step": 816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9830564600285525e-05,
+      "loss": 0.8088,
+      "step": 817
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.983009209449191e-05,
+      "loss": 0.8413,
+      "step": 818
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9829618932646094e-05,
+      "loss": 0.6666,
+      "step": 819
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.982914511475847e-05,
+      "loss": 0.843,
+      "step": 820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.982867064083946e-05,
+      "loss": 0.7472,
+      "step": 821
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.982819551089951e-05,
+      "loss": 0.808,
+      "step": 822
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9827719724949054e-05,
+      "loss": 0.7469,
+      "step": 823
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.982724328299856e-05,
+      "loss": 0.907,
+      "step": 824
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.982676618505851e-05,
+      "loss": 0.7773,
+      "step": 825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.982628843113939e-05,
+      "loss": 0.7778,
+      "step": 826
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.982581002125173e-05,
+      "loss": 0.7446,
+      "step": 827
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.982533095540603e-05,
+      "loss": 0.7757,
+      "step": 828
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.982485123361282e-05,
+      "loss": 0.9022,
+      "step": 829
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9824370855882674e-05,
+      "loss": 0.8788,
+      "step": 830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9823889822226135e-05,
+      "loss": 0.7074,
+      "step": 831
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.982340813265379e-05,
+      "loss": 0.9032,
+      "step": 832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.982292578717625e-05,
+      "loss": 0.7929,
+      "step": 833
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.98224427858041e-05,
+      "loss": 0.7907,
+      "step": 834
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.982195912854796e-05,
+      "loss": 0.7923,
+      "step": 835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.982147481541849e-05,
+      "loss": 0.8463,
+      "step": 836
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.982098984642633e-05,
+      "loss": 0.9368,
+      "step": 837
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.982050422158214e-05,
+      "loss": 0.8556,
+      "step": 838
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.98200179408966e-05,
+      "loss": 0.7811,
+      "step": 839
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.981953100438041e-05,
+      "loss": 0.7605,
+      "step": 840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.981904341204428e-05,
+      "loss": 0.7855,
+      "step": 841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.981855516389894e-05,
+      "loss": 0.7145,
+      "step": 842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.981806625995512e-05,
+      "loss": 0.8521,
+      "step": 843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9817576700223566e-05,
+      "loss": 0.8012,
+      "step": 844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.981708648471506e-05,
+      "loss": 0.7597,
+      "step": 845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.981659561344036e-05,
+      "loss": 0.912,
+      "step": 846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.981610408641029e-05,
+      "loss": 0.8165,
+      "step": 847
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9815611903635647e-05,
+      "loss": 0.8769,
+      "step": 848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.981511906512725e-05,
+      "loss": 0.7305,
+      "step": 849
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9814625570895954e-05,
+      "loss": 0.8208,
+      "step": 850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.98141314209526e-05,
+      "loss": 0.8546,
+      "step": 851
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.981363661530805e-05,
+      "loss": 0.8462,
+      "step": 852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.981314115397319e-05,
+      "loss": 0.8703,
+      "step": 853
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9812645036958935e-05,
+      "loss": 0.7669,
+      "step": 854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.981214826427617e-05,
+      "loss": 0.7912,
+      "step": 855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.981165083593584e-05,
+      "loss": 0.7443,
+      "step": 856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.981115275194888e-05,
+      "loss": 0.8007,
+      "step": 857
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.981065401232624e-05,
+      "loss": 0.7951,
+      "step": 858
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9810154617078884e-05,
+      "loss": 0.8467,
+      "step": 859
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.980965456621781e-05,
+      "loss": 0.8174,
+      "step": 860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9809153859753994e-05,
+      "loss": 0.7647,
+      "step": 861
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.980865249769847e-05,
+      "loss": 0.7395,
+      "step": 862
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.980815048006226e-05,
+      "loss": 0.7605,
+      "step": 863
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.980764780685639e-05,
+      "loss": 0.7465,
+      "step": 864
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.980714447809193e-05,
+      "loss": 0.7715,
+      "step": 865
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.980664049377995e-05,
+      "loss": 0.842,
+      "step": 866
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.980613585393153e-05,
+      "loss": 0.715,
+      "step": 867
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9805630558557755e-05,
+      "loss": 0.765,
+      "step": 868
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9805124607669764e-05,
+      "loss": 0.7833,
+      "step": 869
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.980461800127867e-05,
+      "loss": 0.9165,
+      "step": 870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.980411073939562e-05,
+      "loss": 0.6855,
+      "step": 871
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.980360282203176e-05,
+      "loss": 0.9242,
+      "step": 872
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9803094249198266e-05,
+      "loss": 0.9096,
+      "step": 873
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9802585020906326e-05,
+      "loss": 0.6016,
+      "step": 874
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.980207513716714e-05,
+      "loss": 0.7368,
+      "step": 875
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9801564597991906e-05,
+      "loss": 0.8978,
+      "step": 876
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.980105340339188e-05,
+      "loss": 0.7743,
+      "step": 877
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.980054155337829e-05,
+      "loss": 0.6764,
+      "step": 878
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.980002904796239e-05,
+      "loss": 0.7444,
+      "step": 879
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.979951588715545e-05,
+      "loss": 0.7704,
+      "step": 880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.979900207096877e-05,
+      "loss": 0.8279,
+      "step": 881
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9798487599413644e-05,
+      "loss": 0.7875,
+      "step": 882
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.979797247250137e-05,
+      "loss": 0.6874,
+      "step": 883
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.97974566902433e-05,
+      "loss": 0.8578,
+      "step": 884
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.979694025265077e-05,
+      "loss": 0.8103,
+      "step": 885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.979642315973513e-05,
+      "loss": 0.758,
+      "step": 886
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9795905411507764e-05,
+      "loss": 0.6735,
+      "step": 887
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.979538700798005e-05,
+      "loss": 0.6841,
+      "step": 888
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.97948679491634e-05,
+      "loss": 0.8136,
+      "step": 889
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9794348235069216e-05,
+      "loss": 0.7573,
+      "step": 890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.979382786570894e-05,
+      "loss": 0.6977,
+      "step": 891
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9793306841094005e-05,
+      "loss": 0.7962,
+      "step": 892
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.979278516123588e-05,
+      "loss": 0.8392,
+      "step": 893
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.979226282614604e-05,
+      "loss": 0.7109,
+      "step": 894
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9791739835835957e-05,
+      "loss": 0.7142,
+      "step": 895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9791216190317144e-05,
+      "loss": 0.7309,
+      "step": 896
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.979069188960113e-05,
+      "loss": 0.8165,
+      "step": 897
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.979016693369942e-05,
+      "loss": 0.7918,
+      "step": 898
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9789641322623574e-05,
+      "loss": 0.9063,
+      "step": 899
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9789115056385155e-05,
+      "loss": 0.8208,
+      "step": 900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.978858813499573e-05,
+      "loss": 0.8047,
+      "step": 901
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.978806055846689e-05,
+      "loss": 0.838,
+      "step": 902
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.978753232681023e-05,
+      "loss": 0.5881,
+      "step": 903
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.978700344003738e-05,
+      "loss": 0.72,
+      "step": 904
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9786473898159965e-05,
+      "loss": 0.7764,
+      "step": 905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.978594370118964e-05,
+      "loss": 0.7764,
+      "step": 906
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.978541284913805e-05,
+      "loss": 0.8236,
+      "step": 907
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.978488134201689e-05,
+      "loss": 0.7431,
+      "step": 908
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9784349179837824e-05,
+      "loss": 0.7948,
+      "step": 909
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9783816362612574e-05,
+      "loss": 0.7852,
+      "step": 910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.978328289035285e-05,
+      "loss": 0.8615,
+      "step": 911
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.97827487630704e-05,
+      "loss": 0.7976,
+      "step": 912
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.978221398077695e-05,
+      "loss": 0.7463,
+      "step": 913
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.978167854348428e-05,
+      "loss": 0.7353,
+      "step": 914
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.978114245120415e-05,
+      "loss": 0.7647,
+      "step": 915
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9780605703948356e-05,
+      "loss": 0.8075,
+      "step": 916
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9780068301728714e-05,
+      "loss": 0.8344,
+      "step": 917
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.977953024455702e-05,
+      "loss": 0.8024,
+      "step": 918
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.977899153244513e-05,
+      "loss": 0.7432,
+      "step": 919
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.977845216540488e-05,
+      "loss": 0.6799,
+      "step": 920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.977791214344813e-05,
+      "loss": 0.9288,
+      "step": 921
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9777371466586766e-05,
+      "loss": 0.7701,
+      "step": 922
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.977683013483267e-05,
+      "loss": 0.8723,
+      "step": 923
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.977628814819776e-05,
+      "loss": 0.8407,
+      "step": 924
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.977574550669394e-05,
+      "loss": 0.7723,
+      "step": 925
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.977520221033315e-05,
+      "loss": 0.806,
+      "step": 926
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9774658259127356e-05,
+      "loss": 0.7852,
+      "step": 927
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.977411365308849e-05,
+      "loss": 0.7003,
+      "step": 928
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9773568392228546e-05,
+      "loss": 0.7771,
+      "step": 929
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.977302247655952e-05,
+      "loss": 0.8305,
+      "step": 930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.977247590609341e-05,
+      "loss": 0.8523,
+      "step": 931
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.977192868084225e-05,
+      "loss": 0.7748,
+      "step": 932
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.977138080081806e-05,
+      "loss": 0.8938,
+      "step": 933
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.977083226603289e-05,
+      "loss": 0.8277,
+      "step": 934
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.977028307649882e-05,
+      "loss": 0.6922,
+      "step": 935
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.97697332322279e-05,
+      "loss": 0.7355,
+      "step": 936
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.976918273323225e-05,
+      "loss": 0.835,
+      "step": 937
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9768631579523965e-05,
+      "loss": 0.7136,
+      "step": 938
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.976807977111517e-05,
+      "loss": 0.8039,
+      "step": 939
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9767527308018e-05,
+      "loss": 0.8203,
+      "step": 940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.976697419024461e-05,
+      "loss": 0.7982,
+      "step": 941
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.976642041780714e-05,
+      "loss": 0.8287,
+      "step": 942
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.976586599071781e-05,
+      "loss": 0.7634,
+      "step": 943
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9765310908988775e-05,
+      "loss": 0.7367,
+      "step": 944
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.976475517263227e-05,
+      "loss": 0.8454,
+      "step": 945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9764198781660504e-05,
+      "loss": 0.7745,
+      "step": 946
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9763641736085716e-05,
+      "loss": 0.7066,
+      "step": 947
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.976308403592017e-05,
+      "loss": 0.7423,
+      "step": 948
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.97625256811761e-05,
+      "loss": 0.7692,
+      "step": 949
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.976196667186582e-05,
+      "loss": 0.8042,
+      "step": 950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.97614070080016e-05,
+      "loss": 0.7599,
+      "step": 951
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9760846689595765e-05,
+      "loss": 0.695,
+      "step": 952
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.976028571666063e-05,
+      "loss": 0.8107,
+      "step": 953
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.975972408920852e-05,
+      "loss": 0.7407,
+      "step": 954
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9759161807251817e-05,
+      "loss": 0.7822,
+      "step": 955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.975859887080287e-05,
+      "loss": 0.827,
+      "step": 956
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.975803527987405e-05,
+      "loss": 0.8611,
+      "step": 957
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.975747103447777e-05,
+      "loss": 0.8065,
+      "step": 958
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.975690613462643e-05,
+      "loss": 0.6587,
+      "step": 959
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.975634058033246e-05,
+      "loss": 0.697,
+      "step": 960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9755774371608286e-05,
+      "loss": 0.7989,
+      "step": 961
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9755207508466376e-05,
+      "loss": 0.8024,
+      "step": 962
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9754639990919185e-05,
+      "loss": 0.8347,
+      "step": 963
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.97540718189792e-05,
+      "loss": 0.833,
+      "step": 964
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.975350299265892e-05,
+      "loss": 0.8449,
+      "step": 965
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.975293351197085e-05,
+      "loss": 0.698,
+      "step": 966
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9752363376927506e-05,
+      "loss": 0.7172,
+      "step": 967
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.975179258754144e-05,
+      "loss": 0.7501,
+      "step": 968
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.975122114382521e-05,
+      "loss": 0.782,
+      "step": 969
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.975064904579137e-05,
+      "loss": 0.8544,
+      "step": 970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.97500762934525e-05,
+      "loss": 0.7223,
+      "step": 971
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9749502886821215e-05,
+      "loss": 0.8507,
+      "step": 972
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.974892882591011e-05,
+      "loss": 0.9346,
+      "step": 973
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.974835411073182e-05,
+      "loss": 0.7993,
+      "step": 974
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9747778741298974e-05,
+      "loss": 0.7806,
+      "step": 975
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.974720271762424e-05,
+      "loss": 0.8959,
+      "step": 976
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9746626039720274e-05,
+      "loss": 0.9167,
+      "step": 977
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.974604870759976e-05,
+      "loss": 0.7677,
+      "step": 978
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9745470721275406e-05,
+      "loss": 0.6733,
+      "step": 979
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.974489208075991e-05,
+      "loss": 0.8375,
+      "step": 980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.974431278606601e-05,
+      "loss": 0.8362,
+      "step": 981
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.974373283720643e-05,
+      "loss": 0.8631,
+      "step": 982
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9743152234193954e-05,
+      "loss": 0.756,
+      "step": 983
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9742570977041314e-05,
+      "loss": 0.7069,
+      "step": 984
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.974198906576133e-05,
+      "loss": 0.8407,
+      "step": 985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.974140650036676e-05,
+      "loss": 0.8172,
+      "step": 986
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.974082328087045e-05,
+      "loss": 0.8592,
+      "step": 987
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.974023940728523e-05,
+      "loss": 0.8113,
+      "step": 988
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.973965487962391e-05,
+      "loss": 0.8768,
+      "step": 989
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.973906969789937e-05,
+      "loss": 0.8556,
+      "step": 990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.973848386212446e-05,
+      "loss": 0.8262,
+      "step": 991
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.973789737231208e-05,
+      "loss": 0.7528,
+      "step": 992
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.973731022847513e-05,
+      "loss": 0.9564,
+      "step": 993
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9736722430626516e-05,
+      "loss": 0.7519,
+      "step": 994
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.973613397877917e-05,
+      "loss": 0.7529,
+      "step": 995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.973554487294603e-05,
+      "loss": 0.7365,
+      "step": 996
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.973495511314005e-05,
+      "loss": 0.7971,
+      "step": 997
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.97343646993742e-05,
+      "loss": 0.8946,
+      "step": 998
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.973377363166148e-05,
+      "loss": 0.7756,
+      "step": 999
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.973318191001487e-05,
+      "loss": 0.8715,
+      "step": 1000
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 21284,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 1000,
+  "total_flos": 1.3248648828220867e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}