diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,33415 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 45.0,
+  "global_step": 5490,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 0,
+      "loss": 2.1445,
+      "step": 1
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0,
+      "loss": 1.5811,
+      "step": 2
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0,
+      "loss": 1.8887,
+      "step": 3
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0,
+      "loss": 1.7783,
+      "step": 4
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0,
+      "loss": 1.0605,
+      "step": 5
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0,
+      "loss": 1.7715,
+      "step": 6
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0,
+      "loss": 1.6865,
+      "step": 7
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2920296742201791e-05,
+      "loss": 2.2617,
+      "step": 8
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.0478185834579563e-05,
+      "loss": 3.2812,
+      "step": 9
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5840593484403582e-05,
+      "loss": 1.6172,
+      "step": 10
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3e-05,
+      "loss": 1.0195,
+      "step": 11
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3e-05,
+      "loss": 2.2012,
+      "step": 12
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3e-05,
+      "loss": 2.4473,
+      "step": 13
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3e-05,
+      "loss": 1.8857,
+      "step": 14
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3e-05,
+      "loss": 0.9712,
+      "step": 15
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3e-05,
+      "loss": 1.4219,
+      "step": 16
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3e-05,
+      "loss": 1.4658,
+      "step": 17
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3e-05,
+      "loss": 1.3623,
+      "step": 18
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3e-05,
+      "loss": 0.9082,
+      "step": 19
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3e-05,
+      "loss": 0.7769,
+      "step": 20
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3e-05,
+      "loss": 1.2158,
+      "step": 21
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3e-05,
+      "loss": 1.168,
+      "step": 22
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3e-05,
+      "loss": 0.8076,
+      "step": 23
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3e-05,
+      "loss": 0.7373,
+      "step": 24
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3e-05,
+      "loss": 0.9248,
+      "step": 25
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3e-05,
+      "loss": 0.7666,
+      "step": 26
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3e-05,
+      "loss": 0.9561,
+      "step": 27
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3e-05,
+      "loss": 0.9736,
+      "step": 28
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3e-05,
+      "loss": 0.5562,
+      "step": 29
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3e-05,
+      "loss": 0.6265,
+      "step": 30
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3e-05,
+      "loss": 0.8286,
+      "step": 31
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3e-05,
+      "loss": 0.8613,
+      "step": 32
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3e-05,
+      "loss": 1.0742,
+      "step": 33
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3e-05,
+      "loss": 1.0586,
+      "step": 34
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3e-05,
+      "loss": 0.7891,
+      "step": 35
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3e-05,
+      "loss": 0.6924,
+      "step": 36
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3e-05,
+      "loss": 0.5786,
+      "step": 37
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3e-05,
+      "loss": 0.6812,
+      "step": 38
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3e-05,
+      "loss": 0.6182,
+      "step": 39
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3e-05,
+      "loss": 0.894,
+      "step": 40
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3e-05,
+      "loss": 0.5898,
+      "step": 41
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3e-05,
+      "loss": 0.7534,
+      "step": 42
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3e-05,
+      "loss": 0.646,
+      "step": 43
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3e-05,
+      "loss": 0.8872,
+      "step": 44
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3e-05,
+      "loss": 0.6396,
+      "step": 45
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3e-05,
+      "loss": 0.7437,
+      "step": 46
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3e-05,
+      "loss": 0.623,
+      "step": 47
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3e-05,
+      "loss": 0.5605,
+      "step": 48
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3e-05,
+      "loss": 0.6196,
+      "step": 49
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3e-05,
+      "loss": 0.604,
+      "step": 50
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3e-05,
+      "loss": 1.0996,
+      "step": 51
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3e-05,
+      "loss": 0.7144,
+      "step": 52
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3e-05,
+      "loss": 0.6431,
+      "step": 53
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3e-05,
+      "loss": 0.9429,
+      "step": 54
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3e-05,
+      "loss": 0.6787,
+      "step": 55
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3e-05,
+      "loss": 0.5439,
+      "step": 56
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3e-05,
+      "loss": 0.729,
+      "step": 57
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3e-05,
+      "loss": 0.5762,
+      "step": 58
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3e-05,
+      "loss": 0.6382,
+      "step": 59
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3e-05,
+      "loss": 0.7124,
+      "step": 60
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3e-05,
+      "loss": 0.8062,
+      "step": 61
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3e-05,
+      "loss": 0.6553,
+      "step": 62
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3e-05,
+      "loss": 0.6172,
+      "step": 63
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3e-05,
+      "loss": 1.0049,
+      "step": 64
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3e-05,
+      "loss": 0.811,
+      "step": 65
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3e-05,
+      "loss": 0.4766,
+      "step": 66
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3e-05,
+      "loss": 0.7153,
+      "step": 67
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3e-05,
+      "loss": 0.5312,
+      "step": 68
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3e-05,
+      "loss": 0.5991,
+      "step": 69
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3e-05,
+      "loss": 0.3689,
+      "step": 70
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3e-05,
+      "loss": 0.9146,
+      "step": 71
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3e-05,
+      "loss": 0.603,
+      "step": 72
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3e-05,
+      "loss": 0.5078,
+      "step": 73
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3e-05,
+      "loss": 0.6914,
+      "step": 74
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3e-05,
+      "loss": 0.6455,
+      "step": 75
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3e-05,
+      "loss": 0.687,
+      "step": 76
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3e-05,
+      "loss": 0.5439,
+      "step": 77
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3e-05,
+      "loss": 0.4917,
+      "step": 78
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3e-05,
+      "loss": 0.6025,
+      "step": 79
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3e-05,
+      "loss": 0.6538,
+      "step": 80
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3e-05,
+      "loss": 0.8042,
+      "step": 81
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3e-05,
+      "loss": 0.541,
+      "step": 82
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3e-05,
+      "loss": 0.8457,
+      "step": 83
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3e-05,
+      "loss": 0.5537,
+      "step": 84
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3e-05,
+      "loss": 0.583,
+      "step": 85
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3e-05,
+      "loss": 0.5996,
+      "step": 86
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3e-05,
+      "loss": 0.5967,
+      "step": 87
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3e-05,
+      "loss": 0.6572,
+      "step": 88
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3e-05,
+      "loss": 0.5508,
+      "step": 89
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3e-05,
+      "loss": 0.4355,
+      "step": 90
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3e-05,
+      "loss": 0.6362,
+      "step": 91
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3e-05,
+      "loss": 0.5098,
+      "step": 92
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3e-05,
+      "loss": 0.6787,
+      "step": 93
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3e-05,
+      "loss": 0.5635,
+      "step": 94
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3e-05,
+      "loss": 0.5234,
+      "step": 95
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3e-05,
+      "loss": 0.4456,
+      "step": 96
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3e-05,
+      "loss": 0.5156,
+      "step": 97
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3e-05,
+      "loss": 0.4717,
+      "step": 98
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3e-05,
+      "loss": 0.6074,
+      "step": 99
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3e-05,
+      "loss": 0.7568,
+      "step": 100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3e-05,
+      "loss": 0.6426,
+      "step": 101
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3e-05,
+      "loss": 0.5601,
+      "step": 102
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3e-05,
+      "loss": 0.6499,
+      "step": 103
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3e-05,
+      "loss": 0.8027,
+      "step": 104
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3e-05,
+      "loss": 0.8242,
+      "step": 105
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3e-05,
+      "loss": 0.5366,
+      "step": 106
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3e-05,
+      "loss": 0.46,
+      "step": 107
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3e-05,
+      "loss": 0.5312,
+      "step": 108
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3e-05,
+      "loss": 0.4724,
+      "step": 109
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3e-05,
+      "loss": 0.4373,
+      "step": 110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3e-05,
+      "loss": 0.6592,
+      "step": 111
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3e-05,
+      "loss": 0.5425,
+      "step": 112
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3e-05,
+      "loss": 0.6465,
+      "step": 113
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3e-05,
+      "loss": 0.4504,
+      "step": 114
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3e-05,
+      "loss": 0.4641,
+      "step": 115
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3e-05,
+      "loss": 0.4351,
+      "step": 116
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3e-05,
+      "loss": 0.5239,
+      "step": 117
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3e-05,
+      "loss": 0.8569,
+      "step": 118
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3e-05,
+      "loss": 0.604,
+      "step": 119
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3e-05,
+      "loss": 0.7881,
+      "step": 120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3e-05,
+      "loss": 0.7637,
+      "step": 121
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3e-05,
+      "loss": 0.252,
+      "step": 122
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.59912109375,
+      "eval_mae": 0.6083984375,
+      "eval_mse": 0.59912109375,
+      "eval_runtime": 51.0188,
+      "eval_samples_per_second": 3241.806,
+      "eval_steps_per_second": 0.274,
+      "step": 122
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3e-05,
+      "loss": 0.7241,
+      "step": 123
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3e-05,
+      "loss": 0.542,
+      "step": 124
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3e-05,
+      "loss": 0.4163,
+      "step": 125
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3e-05,
+      "loss": 0.6753,
+      "step": 126
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3e-05,
+      "loss": 0.5562,
+      "step": 127
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3e-05,
+      "loss": 0.5186,
+      "step": 128
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3e-05,
+      "loss": 0.7119,
+      "step": 129
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3e-05,
+      "loss": 0.8906,
+      "step": 130
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3e-05,
+      "loss": 0.5962,
+      "step": 131
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3e-05,
+      "loss": 0.4604,
+      "step": 132
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3e-05,
+      "loss": 0.6694,
+      "step": 133
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3e-05,
+      "loss": 0.5757,
+      "step": 134
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3e-05,
+      "loss": 0.5654,
+      "step": 135
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3e-05,
+      "loss": 0.6099,
+      "step": 136
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3e-05,
+      "loss": 0.7466,
+      "step": 137
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3e-05,
+      "loss": 0.7339,
+      "step": 138
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3e-05,
+      "loss": 0.6538,
+      "step": 139
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3e-05,
+      "loss": 0.5171,
+      "step": 140
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3e-05,
+      "loss": 0.4995,
+      "step": 141
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3e-05,
+      "loss": 0.5967,
+      "step": 142
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3e-05,
+      "loss": 0.5171,
+      "step": 143
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3e-05,
+      "loss": 0.7134,
+      "step": 144
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3e-05,
+      "loss": 0.5039,
+      "step": 145
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3e-05,
+      "loss": 0.6567,
+      "step": 146
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3e-05,
+      "loss": 0.5889,
+      "step": 147
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3e-05,
+      "loss": 0.7588,
+      "step": 148
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3e-05,
+      "loss": 0.5928,
+      "step": 149
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3e-05,
+      "loss": 0.5029,
+      "step": 150
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3e-05,
+      "loss": 0.5488,
+      "step": 151
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3e-05,
+      "loss": 0.3921,
+      "step": 152
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3e-05,
+      "loss": 0.7207,
+      "step": 153
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3e-05,
+      "loss": 0.5352,
+      "step": 154
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3e-05,
+      "loss": 0.5371,
+      "step": 155
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3e-05,
+      "loss": 0.6509,
+      "step": 156
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3e-05,
+      "loss": 0.5649,
+      "step": 157
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3e-05,
+      "loss": 0.748,
+      "step": 158
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3e-05,
+      "loss": 0.542,
+      "step": 159
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3e-05,
+      "loss": 0.5454,
+      "step": 160
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3e-05,
+      "loss": 0.6011,
+      "step": 161
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3e-05,
+      "loss": 0.5171,
+      "step": 162
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3e-05,
+      "loss": 0.4204,
+      "step": 163
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3e-05,
+      "loss": 0.4995,
+      "step": 164
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3e-05,
+      "loss": 0.6279,
+      "step": 165
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3e-05,
+      "loss": 0.603,
+      "step": 166
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3e-05,
+      "loss": 0.5693,
+      "step": 167
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3e-05,
+      "loss": 0.4265,
+      "step": 168
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3e-05,
+      "loss": 0.542,
+      "step": 169
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3e-05,
+      "loss": 0.7119,
+      "step": 170
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3e-05,
+      "loss": 0.406,
+      "step": 171
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3e-05,
+      "loss": 0.5874,
+      "step": 172
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3e-05,
+      "loss": 0.6558,
+      "step": 173
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3e-05,
+      "loss": 0.54,
+      "step": 174
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3e-05,
+      "loss": 0.7646,
+      "step": 175
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3e-05,
+      "loss": 0.5142,
+      "step": 176
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3e-05,
+      "loss": 0.5386,
+      "step": 177
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3e-05,
+      "loss": 0.688,
+      "step": 178
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3e-05,
+      "loss": 0.7114,
+      "step": 179
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3e-05,
+      "loss": 0.4976,
+      "step": 180
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3e-05,
+      "loss": 0.543,
+      "step": 181
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3e-05,
+      "loss": 0.6089,
+      "step": 182
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3e-05,
+      "loss": 0.6084,
+      "step": 183
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3e-05,
+      "loss": 0.4351,
+      "step": 184
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3e-05,
+      "loss": 0.6606,
+      "step": 185
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3e-05,
+      "loss": 0.6465,
+      "step": 186
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3e-05,
+      "loss": 0.4043,
+      "step": 187
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3e-05,
+      "loss": 0.4106,
+      "step": 188
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3e-05,
+      "loss": 0.7227,
+      "step": 189
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3e-05,
+      "loss": 0.45,
+      "step": 190
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3e-05,
+      "loss": 0.4104,
+      "step": 191
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3e-05,
+      "loss": 0.5137,
+      "step": 192
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3e-05,
+      "loss": 0.3804,
+      "step": 193
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3e-05,
+      "loss": 0.4985,
+      "step": 194
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3e-05,
+      "loss": 0.4133,
+      "step": 195
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3e-05,
+      "loss": 0.6768,
+      "step": 196
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3e-05,
+      "loss": 0.4177,
+      "step": 197
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3e-05,
+      "loss": 0.5161,
+      "step": 198
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3e-05,
+      "loss": 0.4924,
+      "step": 199
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3e-05,
+      "loss": 0.354,
+      "step": 200
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3e-05,
+      "loss": 0.5913,
+      "step": 201
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3e-05,
+      "loss": 0.521,
+      "step": 202
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3e-05,
+      "loss": 0.9839,
+      "step": 203
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3e-05,
+      "loss": 0.4277,
+      "step": 204
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3e-05,
+      "loss": 0.6826,
+      "step": 205
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3e-05,
+      "loss": 0.3484,
+      "step": 206
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3e-05,
+      "loss": 0.4324,
+      "step": 207
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3e-05,
+      "loss": 0.4888,
+      "step": 208
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3e-05,
+      "loss": 0.3879,
+      "step": 209
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3e-05,
+      "loss": 0.4697,
+      "step": 210
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3e-05,
+      "loss": 0.7661,
+      "step": 211
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3e-05,
+      "loss": 0.4175,
+      "step": 212
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3e-05,
+      "loss": 0.4832,
+      "step": 213
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3e-05,
+      "loss": 0.5459,
+      "step": 214
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3e-05,
+      "loss": 0.4927,
+      "step": 215
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3e-05,
+      "loss": 0.5264,
+      "step": 216
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3e-05,
+      "loss": 0.688,
+      "step": 217
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3e-05,
+      "loss": 0.4841,
+      "step": 218
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3e-05,
+      "loss": 0.5728,
+      "step": 219
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3e-05,
+      "loss": 0.3945,
+      "step": 220
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3e-05,
+      "loss": 0.5361,
+      "step": 221
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3e-05,
+      "loss": 0.5742,
+      "step": 222
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3e-05,
+      "loss": 0.627,
+      "step": 223
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3e-05,
+      "loss": 0.4314,
+      "step": 224
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3e-05,
+      "loss": 0.5186,
+      "step": 225
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3e-05,
+      "loss": 0.4697,
+      "step": 226
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3e-05,
+      "loss": 0.6445,
+      "step": 227
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3e-05,
+      "loss": 0.4043,
+      "step": 228
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3e-05,
+      "loss": 0.4126,
+      "step": 229
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3e-05,
+      "loss": 0.5576,
+      "step": 230
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3e-05,
+      "loss": 0.5234,
+      "step": 231
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3e-05,
+      "loss": 0.5518,
+      "step": 232
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3e-05,
+      "loss": 0.686,
+      "step": 233
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3e-05,
+      "loss": 0.522,
+      "step": 234
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3e-05,
+      "loss": 0.4934,
+      "step": 235
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3e-05,
+      "loss": 0.436,
+      "step": 236
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3e-05,
+      "loss": 0.4109,
+      "step": 237
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3e-05,
+      "loss": 0.6128,
+      "step": 238
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3e-05,
+      "loss": 0.4937,
+      "step": 239
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3e-05,
+      "loss": 0.5928,
+      "step": 240
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3e-05,
+      "loss": 0.408,
+      "step": 241
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3e-05,
+      "loss": 0.5083,
+      "step": 242
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3e-05,
+      "loss": 0.52,
+      "step": 243
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3e-05,
+      "loss": 0.52,
+      "step": 244
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 0.5244140625,
+      "eval_mae": 0.564453125,
+      "eval_mse": 0.5244140625,
+      "eval_runtime": 51.1669,
+      "eval_samples_per_second": 3232.424,
+      "eval_steps_per_second": 0.274,
+      "step": 244
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3e-05,
+      "loss": 0.6333,
+      "step": 245
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3e-05,
+      "loss": 0.7544,
+      "step": 246
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3e-05,
+      "loss": 0.5146,
+      "step": 247
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3e-05,
+      "loss": 0.8477,
+      "step": 248
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3e-05,
+      "loss": 0.5869,
+      "step": 249
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3e-05,
+      "loss": 0.4473,
+      "step": 250
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3e-05,
+      "loss": 0.5854,
+      "step": 251
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3e-05,
+      "loss": 0.5981,
+      "step": 252
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3e-05,
+      "loss": 0.5527,
+      "step": 253
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3e-05,
+      "loss": 0.4861,
+      "step": 254
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3e-05,
+      "loss": 0.4646,
+      "step": 255
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3e-05,
+      "loss": 0.5259,
+      "step": 256
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3e-05,
+      "loss": 0.4365,
+      "step": 257
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3e-05,
+      "loss": 0.4983,
+      "step": 258
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3e-05,
+      "loss": 0.4792,
+      "step": 259
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3e-05,
+      "loss": 1.0342,
+      "step": 260
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3e-05,
+      "loss": 0.5503,
+      "step": 261
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3e-05,
+      "loss": 0.6089,
+      "step": 262
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3e-05,
+      "loss": 0.4309,
+      "step": 263
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3e-05,
+      "loss": 0.5684,
+      "step": 264
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3e-05,
+      "loss": 0.7822,
+      "step": 265
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3e-05,
+      "loss": 0.3884,
+      "step": 266
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3e-05,
+      "loss": 0.3989,
+      "step": 267
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3e-05,
+      "loss": 0.5986,
+      "step": 268
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3e-05,
+      "loss": 0.4829,
+      "step": 269
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3e-05,
+      "loss": 0.519,
+      "step": 270
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3e-05,
+      "loss": 0.8662,
+      "step": 271
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3e-05,
+      "loss": 0.4707,
+      "step": 272
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3e-05,
+      "loss": 0.448,
+      "step": 273
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 0.4294,
+      "step": 274
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 0.4751,
+      "step": 275
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3e-05,
+      "loss": 1.0469,
+      "step": 276
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3e-05,
+      "loss": 0.3572,
+      "step": 277
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3e-05,
+      "loss": 0.4182,
+      "step": 278
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3e-05,
+      "loss": 0.4514,
+      "step": 279
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3e-05,
+      "loss": 0.561,
+      "step": 280
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3e-05,
+      "loss": 0.3677,
+      "step": 281
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3e-05,
+      "loss": 0.7056,
+      "step": 282
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3e-05,
+      "loss": 0.5381,
+      "step": 283
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3e-05,
+      "loss": 0.6025,
+      "step": 284
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3e-05,
+      "loss": 0.3469,
+      "step": 285
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3e-05,
+      "loss": 0.4685,
+      "step": 286
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3e-05,
+      "loss": 0.5815,
+      "step": 287
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3e-05,
+      "loss": 0.5039,
+      "step": 288
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3e-05,
+      "loss": 0.5059,
+      "step": 289
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3e-05,
+      "loss": 0.5015,
+      "step": 290
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3e-05,
+      "loss": 0.4739,
+      "step": 291
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3e-05,
+      "loss": 0.6128,
+      "step": 292
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3e-05,
+      "loss": 0.4734,
+      "step": 293
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3e-05,
+      "loss": 0.4597,
+      "step": 294
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3e-05,
+      "loss": 0.3843,
+      "step": 295
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3e-05,
+      "loss": 0.562,
+      "step": 296
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3e-05,
+      "loss": 0.6689,
+      "step": 297
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3e-05,
+      "loss": 0.7383,
+      "step": 298
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3e-05,
+      "loss": 0.4624,
+      "step": 299
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3e-05,
+      "loss": 0.3574,
+      "step": 300
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3e-05,
+      "loss": 0.4661,
+      "step": 301
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3e-05,
+      "loss": 0.3926,
+      "step": 302
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3e-05,
+      "loss": 0.667,
+      "step": 303
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3e-05,
+      "loss": 0.4175,
+      "step": 304
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 0.7905,
+      "step": 305
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3e-05,
+      "loss": 0.6289,
+      "step": 306
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3e-05,
+      "loss": 0.4531,
+      "step": 307
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3e-05,
+      "loss": 0.603,
+      "step": 308
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3e-05,
+      "loss": 0.5039,
+      "step": 309
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3e-05,
+      "loss": 0.4744,
+      "step": 310
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3e-05,
+      "loss": 0.3943,
+      "step": 311
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3e-05,
+      "loss": 0.6084,
+      "step": 312
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3e-05,
+      "loss": 0.4622,
+      "step": 313
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3e-05,
+      "loss": 0.5225,
+      "step": 314
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3e-05,
+      "loss": 0.5132,
+      "step": 315
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3e-05,
+      "loss": 0.3008,
+      "step": 316
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3e-05,
+      "loss": 0.5742,
+      "step": 317
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3e-05,
+      "loss": 0.3984,
+      "step": 318
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3e-05,
+      "loss": 0.5513,
+      "step": 319
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3e-05,
+      "loss": 0.4075,
+      "step": 320
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3e-05,
+      "loss": 0.6548,
+      "step": 321
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3e-05,
+      "loss": 0.4075,
+      "step": 322
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3e-05,
+      "loss": 0.5005,
+      "step": 323
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3e-05,
+      "loss": 0.4045,
+      "step": 324
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3e-05,
+      "loss": 0.52,
+      "step": 325
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3e-05,
+      "loss": 0.7363,
+      "step": 326
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3e-05,
+      "loss": 0.3452,
+      "step": 327
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3e-05,
+      "loss": 0.4871,
+      "step": 328
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3e-05,
+      "loss": 0.4744,
+      "step": 329
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3e-05,
+      "loss": 0.4619,
+      "step": 330
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3e-05,
+      "loss": 0.4524,
+      "step": 331
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3e-05,
+      "loss": 0.5942,
+      "step": 332
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3e-05,
+      "loss": 0.6978,
+      "step": 333
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3e-05,
+      "loss": 0.5181,
+      "step": 334
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 0.5083,
+      "step": 335
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 0.5181,
+      "step": 336
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3e-05,
+      "loss": 0.3721,
+      "step": 337
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3e-05,
+      "loss": 0.4976,
+      "step": 338
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3e-05,
+      "loss": 0.5625,
+      "step": 339
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3e-05,
+      "loss": 0.4226,
+      "step": 340
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3e-05,
+      "loss": 0.4717,
+      "step": 341
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3e-05,
+      "loss": 0.4675,
+      "step": 342
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3e-05,
+      "loss": 0.3926,
+      "step": 343
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3e-05,
+      "loss": 0.5024,
+      "step": 344
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3e-05,
+      "loss": 0.6362,
+      "step": 345
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3e-05,
+      "loss": 0.3918,
+      "step": 346
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3e-05,
+      "loss": 0.5635,
+      "step": 347
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3e-05,
+      "loss": 0.6108,
+      "step": 348
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3e-05,
+      "loss": 0.4175,
+      "step": 349
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3e-05,
+      "loss": 0.4905,
+      "step": 350
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3e-05,
+      "loss": 0.3906,
+      "step": 351
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3e-05,
+      "loss": 0.4875,
+      "step": 352
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3e-05,
+      "loss": 0.3708,
+      "step": 353
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3e-05,
+      "loss": 0.3682,
+      "step": 354
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3e-05,
+      "loss": 0.5742,
+      "step": 355
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3e-05,
+      "loss": 0.5601,
+      "step": 356
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3e-05,
+      "loss": 0.5303,
+      "step": 357
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3e-05,
+      "loss": 0.46,
+      "step": 358
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3e-05,
+      "loss": 0.3518,
+      "step": 359
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3e-05,
+      "loss": 0.4968,
+      "step": 360
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3e-05,
+      "loss": 0.4258,
+      "step": 361
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3e-05,
+      "loss": 0.4155,
+      "step": 362
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3e-05,
+      "loss": 0.499,
+      "step": 363
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3e-05,
+      "loss": 0.4858,
+      "step": 364
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3e-05,
+      "loss": 0.4155,
+      "step": 365
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 0.71,
+      "step": 366
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 0.479248046875,
+      "eval_mae": 0.5361328125,
+      "eval_mse": 0.479248046875,
+      "eval_runtime": 51.1452,
+      "eval_samples_per_second": 3233.794,
+      "eval_steps_per_second": 0.274,
+      "step": 366
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3e-05,
+      "loss": 0.3301,
+      "step": 367
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3e-05,
+      "loss": 0.4473,
+      "step": 368
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3e-05,
+      "loss": 0.5361,
+      "step": 369
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3e-05,
+      "loss": 0.5415,
+      "step": 370
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3e-05,
+      "loss": 0.689,
+      "step": 371
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3e-05,
+      "loss": 0.4539,
+      "step": 372
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3e-05,
+      "loss": 0.4827,
+      "step": 373
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3e-05,
+      "loss": 0.542,
+      "step": 374
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3e-05,
+      "loss": 0.4299,
+      "step": 375
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3e-05,
+      "loss": 0.4597,
+      "step": 376
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3e-05,
+      "loss": 0.4153,
+      "step": 377
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3e-05,
+      "loss": 0.4463,
+      "step": 378
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3e-05,
+      "loss": 0.5835,
+      "step": 379
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3e-05,
+      "loss": 0.46,
+      "step": 380
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3e-05,
+      "loss": 0.4219,
+      "step": 381
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3e-05,
+      "loss": 0.4194,
+      "step": 382
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3e-05,
+      "loss": 0.3979,
+      "step": 383
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3e-05,
+      "loss": 0.4084,
+      "step": 384
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3e-05,
+      "loss": 0.415,
+      "step": 385
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3e-05,
+      "loss": 0.5034,
+      "step": 386
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3e-05,
+      "loss": 0.4814,
+      "step": 387
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3e-05,
+      "loss": 0.4268,
+      "step": 388
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3e-05,
+      "loss": 0.4556,
+      "step": 389
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3e-05,
+      "loss": 0.5811,
+      "step": 390
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3e-05,
+      "loss": 0.4846,
+      "step": 391
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3e-05,
+      "loss": 0.5337,
+      "step": 392
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3e-05,
+      "loss": 0.3533,
+      "step": 393
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3e-05,
+      "loss": 0.6904,
+      "step": 394
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3e-05,
+      "loss": 0.4946,
+      "step": 395
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3e-05,
+      "loss": 0.3843,
+      "step": 396
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3e-05,
+      "loss": 0.4021,
+      "step": 397
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3e-05,
+      "loss": 0.2654,
+      "step": 398
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3e-05,
+      "loss": 0.3687,
+      "step": 399
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3e-05,
+      "loss": 0.7051,
+      "step": 400
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3e-05,
+      "loss": 0.3225,
+      "step": 401
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3e-05,
+      "loss": 0.3142,
+      "step": 402
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3e-05,
+      "loss": 0.666,
+      "step": 403
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3e-05,
+      "loss": 0.5649,
+      "step": 404
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3e-05,
+      "loss": 0.3992,
+      "step": 405
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3e-05,
+      "loss": 0.5493,
+      "step": 406
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3e-05,
+      "loss": 0.3516,
+      "step": 407
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3e-05,
+      "loss": 0.5161,
+      "step": 408
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3e-05,
+      "loss": 0.4255,
+      "step": 409
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3e-05,
+      "loss": 0.4888,
+      "step": 410
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3e-05,
+      "loss": 0.3574,
+      "step": 411
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3e-05,
+      "loss": 0.363,
+      "step": 412
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3e-05,
+      "loss": 0.3313,
+      "step": 413
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3e-05,
+      "loss": 0.4045,
+      "step": 414
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3e-05,
+      "loss": 0.3882,
+      "step": 415
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3e-05,
+      "loss": 0.3572,
+      "step": 416
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3e-05,
+      "loss": 0.324,
+      "step": 417
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3e-05,
+      "loss": 0.3098,
+      "step": 418
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3e-05,
+      "loss": 0.4775,
+      "step": 419
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3e-05,
+      "loss": 0.3682,
+      "step": 420
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3e-05,
+      "loss": 0.5059,
+      "step": 421
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3e-05,
+      "loss": 0.6865,
+      "step": 422
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3e-05,
+      "loss": 0.6021,
+      "step": 423
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3e-05,
+      "loss": 0.3669,
+      "step": 424
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3e-05,
+      "loss": 0.408,
+      "step": 425
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3e-05,
+      "loss": 0.4949,
+      "step": 426
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3e-05,
+      "loss": 0.7886,
+      "step": 427
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3e-05,
+      "loss": 0.4197,
+      "step": 428
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3e-05,
+      "loss": 0.3728,
+      "step": 429
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3e-05,
+      "loss": 0.3733,
+      "step": 430
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3e-05,
+      "loss": 0.5625,
+      "step": 431
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3e-05,
+      "loss": 0.363,
+      "step": 432
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3e-05,
+      "loss": 0.3755,
+      "step": 433
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3e-05,
+      "loss": 0.3162,
+      "step": 434
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3e-05,
+      "loss": 0.605,
+      "step": 435
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3e-05,
+      "loss": 0.2742,
+      "step": 436
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3e-05,
+      "loss": 0.377,
+      "step": 437
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3e-05,
+      "loss": 0.4565,
+      "step": 438
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3e-05,
+      "loss": 0.4182,
+      "step": 439
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3e-05,
+      "loss": 0.4221,
+      "step": 440
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3e-05,
+      "loss": 0.4927,
+      "step": 441
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3e-05,
+      "loss": 0.4492,
+      "step": 442
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3e-05,
+      "loss": 0.4988,
+      "step": 443
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3e-05,
+      "loss": 0.3533,
+      "step": 444
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3e-05,
+      "loss": 0.4028,
+      "step": 445
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3e-05,
+      "loss": 0.303,
+      "step": 446
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3e-05,
+      "loss": 0.3523,
+      "step": 447
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3e-05,
+      "loss": 0.3416,
+      "step": 448
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3e-05,
+      "loss": 0.5361,
+      "step": 449
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3e-05,
+      "loss": 0.4849,
+      "step": 450
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3e-05,
+      "loss": 0.3813,
+      "step": 451
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3e-05,
+      "loss": 0.3506,
+      "step": 452
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3e-05,
+      "loss": 0.3208,
+      "step": 453
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3e-05,
+      "loss": 0.2681,
+      "step": 454
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3e-05,
+      "loss": 0.3884,
+      "step": 455
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3e-05,
+      "loss": 0.6001,
+      "step": 456
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3e-05,
+      "loss": 0.417,
+      "step": 457
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3e-05,
+      "loss": 0.5181,
+      "step": 458
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3e-05,
+      "loss": 0.4158,
+      "step": 459
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3e-05,
+      "loss": 0.4639,
+      "step": 460
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3e-05,
+      "loss": 0.448,
+      "step": 461
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3e-05,
+      "loss": 0.3799,
+      "step": 462
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3e-05,
+      "loss": 0.2942,
+      "step": 463
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3e-05,
+      "loss": 0.3757,
+      "step": 464
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3e-05,
+      "loss": 0.4604,
+      "step": 465
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3e-05,
+      "loss": 0.4041,
+      "step": 466
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3e-05,
+      "loss": 0.4084,
+      "step": 467
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3e-05,
+      "loss": 0.3489,
+      "step": 468
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3e-05,
+      "loss": 0.4058,
+      "step": 469
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3e-05,
+      "loss": 0.4646,
+      "step": 470
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3e-05,
+      "loss": 0.6606,
+      "step": 471
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3e-05,
+      "loss": 0.4583,
+      "step": 472
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3e-05,
+      "loss": 0.3276,
+      "step": 473
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3e-05,
+      "loss": 0.5195,
+      "step": 474
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3e-05,
+      "loss": 0.3274,
+      "step": 475
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3e-05,
+      "loss": 0.426,
+      "step": 476
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3e-05,
+      "loss": 0.3816,
+      "step": 477
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3e-05,
+      "loss": 0.3945,
+      "step": 478
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3e-05,
+      "loss": 0.4619,
+      "step": 479
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3e-05,
+      "loss": 0.4141,
+      "step": 480
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3e-05,
+      "loss": 0.2891,
+      "step": 481
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3e-05,
+      "loss": 0.344,
+      "step": 482
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3e-05,
+      "loss": 0.3303,
+      "step": 483
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3e-05,
+      "loss": 0.3918,
+      "step": 484
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3e-05,
+      "loss": 0.491,
+      "step": 485
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3e-05,
+      "loss": 0.5591,
+      "step": 486
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3e-05,
+      "loss": 0.384,
+      "step": 487
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3e-05,
+      "loss": 0.9497,
+      "step": 488
+    },
+    {
+      "epoch": 4.0,
+      "eval_loss": 0.441650390625,
+      "eval_mae": 0.51220703125,
+      "eval_mse": 0.441650390625,
+      "eval_runtime": 51.1113,
+      "eval_samples_per_second": 3235.94,
+      "eval_steps_per_second": 0.274,
+      "step": 488
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3e-05,
+      "loss": 0.3083,
+      "step": 489
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3e-05,
+      "loss": 0.5249,
+      "step": 490
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3e-05,
+      "loss": 0.5586,
+      "step": 491
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3e-05,
+      "loss": 0.4458,
+      "step": 492
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3e-05,
+      "loss": 0.3906,
+      "step": 493
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3e-05,
+      "loss": 0.4512,
+      "step": 494
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3e-05,
+      "loss": 0.488,
+      "step": 495
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3e-05,
+      "loss": 0.3079,
+      "step": 496
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3e-05,
+      "loss": 0.5205,
+      "step": 497
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3e-05,
+      "loss": 0.3801,
+      "step": 498
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3e-05,
+      "loss": 0.418,
+      "step": 499
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3e-05,
+      "loss": 0.3093,
+      "step": 500
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3e-05,
+      "loss": 0.5283,
+      "step": 501
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3e-05,
+      "loss": 0.415,
+      "step": 502
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3e-05,
+      "loss": 0.4402,
+      "step": 503
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3e-05,
+      "loss": 0.5029,
+      "step": 504
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3e-05,
+      "loss": 0.4519,
+      "step": 505
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3e-05,
+      "loss": 0.6172,
+      "step": 506
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3e-05,
+      "loss": 0.4373,
+      "step": 507
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3e-05,
+      "loss": 0.4861,
+      "step": 508
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3e-05,
+      "loss": 0.4302,
+      "step": 509
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3e-05,
+      "loss": 0.7236,
+      "step": 510
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3e-05,
+      "loss": 0.3184,
+      "step": 511
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3e-05,
+      "loss": 0.4653,
+      "step": 512
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3e-05,
+      "loss": 0.5264,
+      "step": 513
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3e-05,
+      "loss": 0.5356,
+      "step": 514
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3e-05,
+      "loss": 0.3977,
+      "step": 515
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3e-05,
+      "loss": 0.5386,
+      "step": 516
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3e-05,
+      "loss": 0.4839,
+      "step": 517
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3e-05,
+      "loss": 0.3162,
+      "step": 518
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3e-05,
+      "loss": 0.4858,
+      "step": 519
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 3e-05,
+      "loss": 0.3901,
+      "step": 520
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 3e-05,
+      "loss": 0.4336,
+      "step": 521
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3e-05,
+      "loss": 0.4443,
+      "step": 522
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3e-05,
+      "loss": 0.3359,
+      "step": 523
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3e-05,
+      "loss": 0.4382,
+      "step": 524
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3e-05,
+      "loss": 0.3376,
+      "step": 525
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3e-05,
+      "loss": 0.3765,
+      "step": 526
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3e-05,
+      "loss": 0.6191,
+      "step": 527
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3e-05,
+      "loss": 0.3359,
+      "step": 528
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3e-05,
+      "loss": 0.4221,
+      "step": 529
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3e-05,
+      "loss": 0.9238,
+      "step": 530
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3e-05,
+      "loss": 0.4355,
+      "step": 531
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3e-05,
+      "loss": 0.4348,
+      "step": 532
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3e-05,
+      "loss": 0.427,
+      "step": 533
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3e-05,
+      "loss": 0.3032,
+      "step": 534
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3e-05,
+      "loss": 0.3345,
+      "step": 535
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3e-05,
+      "loss": 0.3931,
+      "step": 536
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3e-05,
+      "loss": 0.4504,
+      "step": 537
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3e-05,
+      "loss": 0.4595,
+      "step": 538
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3e-05,
+      "loss": 0.5566,
+      "step": 539
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3e-05,
+      "loss": 0.3167,
+      "step": 540
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3e-05,
+      "loss": 0.2759,
+      "step": 541
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3e-05,
+      "loss": 0.4785,
+      "step": 542
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3e-05,
+      "loss": 0.4387,
+      "step": 543
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3e-05,
+      "loss": 0.4619,
+      "step": 544
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3e-05,
+      "loss": 0.3516,
+      "step": 545
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3e-05,
+      "loss": 0.3574,
+      "step": 546
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3e-05,
+      "loss": 0.3391,
+      "step": 547
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3e-05,
+      "loss": 0.3867,
+      "step": 548
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3e-05,
+      "loss": 0.3987,
+      "step": 549
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3e-05,
+      "loss": 0.5952,
+      "step": 550
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3e-05,
+      "loss": 0.3193,
+      "step": 551
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3e-05,
+      "loss": 0.4568,
+      "step": 552
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3e-05,
+      "loss": 0.4802,
+      "step": 553
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3e-05,
+      "loss": 0.3474,
+      "step": 554
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3e-05,
+      "loss": 0.5205,
+      "step": 555
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3e-05,
+      "loss": 0.3459,
+      "step": 556
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3e-05,
+      "loss": 0.417,
+      "step": 557
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3e-05,
+      "loss": 0.2238,
+      "step": 558
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3e-05,
+      "loss": 0.3826,
+      "step": 559
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3e-05,
+      "loss": 0.3958,
+      "step": 560
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3e-05,
+      "loss": 0.2988,
+      "step": 561
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3e-05,
+      "loss": 0.4788,
+      "step": 562
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3e-05,
+      "loss": 0.3411,
+      "step": 563
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3e-05,
+      "loss": 0.6284,
+      "step": 564
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3e-05,
+      "loss": 0.3381,
+      "step": 565
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3e-05,
+      "loss": 0.4226,
+      "step": 566
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3e-05,
+      "loss": 0.3865,
+      "step": 567
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3e-05,
+      "loss": 0.4409,
+      "step": 568
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3e-05,
+      "loss": 0.5024,
+      "step": 569
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3e-05,
+      "loss": 0.3291,
+      "step": 570
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3e-05,
+      "loss": 0.354,
+      "step": 571
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3e-05,
+      "loss": 0.3091,
+      "step": 572
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3e-05,
+      "loss": 0.2544,
+      "step": 573
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3e-05,
+      "loss": 0.4604,
+      "step": 574
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3e-05,
+      "loss": 0.3408,
+      "step": 575
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3e-05,
+      "loss": 0.3105,
+      "step": 576
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3e-05,
+      "loss": 0.4131,
+      "step": 577
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3e-05,
+      "loss": 0.3884,
+      "step": 578
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3e-05,
+      "loss": 0.2744,
+      "step": 579
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3e-05,
+      "loss": 0.4436,
+      "step": 580
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3e-05,
+      "loss": 0.4795,
+      "step": 581
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3e-05,
+      "loss": 0.4353,
+      "step": 582
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3e-05,
+      "loss": 0.4717,
+      "step": 583
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3e-05,
+      "loss": 0.2761,
+      "step": 584
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3e-05,
+      "loss": 0.4966,
+      "step": 585
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3e-05,
+      "loss": 0.3391,
+      "step": 586
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3e-05,
+      "loss": 0.3594,
+      "step": 587
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3e-05,
+      "loss": 0.3694,
+      "step": 588
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3e-05,
+      "loss": 0.5405,
+      "step": 589
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3e-05,
+      "loss": 0.5029,
+      "step": 590
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3e-05,
+      "loss": 0.3738,
+      "step": 591
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3e-05,
+      "loss": 0.4429,
+      "step": 592
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3e-05,
+      "loss": 0.3586,
+      "step": 593
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3e-05,
+      "loss": 0.3931,
+      "step": 594
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3e-05,
+      "loss": 0.394,
+      "step": 595
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3e-05,
+      "loss": 0.502,
+      "step": 596
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3e-05,
+      "loss": 0.3264,
+      "step": 597
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3e-05,
+      "loss": 0.3281,
+      "step": 598
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3e-05,
+      "loss": 0.4023,
+      "step": 599
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3e-05,
+      "loss": 0.3938,
+      "step": 600
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3e-05,
+      "loss": 0.6597,
+      "step": 601
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3e-05,
+      "loss": 0.3794,
+      "step": 602
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3e-05,
+      "loss": 0.3489,
+      "step": 603
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3e-05,
+      "loss": 0.2988,
+      "step": 604
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3e-05,
+      "loss": 0.4954,
+      "step": 605
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3e-05,
+      "loss": 0.748,
+      "step": 606
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3e-05,
+      "loss": 0.4119,
+      "step": 607
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3e-05,
+      "loss": 0.394,
+      "step": 608
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3e-05,
+      "loss": 0.501,
+      "step": 609
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3e-05,
+      "loss": 0.4285,
+      "step": 610
+    },
+    {
+      "epoch": 5.0,
+      "eval_loss": 0.430419921875,
+      "eval_mae": 0.50390625,
+      "eval_mse": 0.430419921875,
+      "eval_runtime": 51.1097,
+      "eval_samples_per_second": 3236.042,
+      "eval_steps_per_second": 0.274,
+      "step": 610
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3e-05,
+      "loss": 0.356,
+      "step": 611
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3e-05,
+      "loss": 0.3867,
+      "step": 612
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3e-05,
+      "loss": 0.5269,
+      "step": 613
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3e-05,
+      "loss": 0.4116,
+      "step": 614
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3e-05,
+      "loss": 0.5234,
+      "step": 615
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3e-05,
+      "loss": 0.5723,
+      "step": 616
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3e-05,
+      "loss": 0.521,
+      "step": 617
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3e-05,
+      "loss": 0.3865,
+      "step": 618
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3e-05,
+      "loss": 0.4309,
+      "step": 619
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3e-05,
+      "loss": 0.3391,
+      "step": 620
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3e-05,
+      "loss": 0.3647,
+      "step": 621
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3e-05,
+      "loss": 0.469,
+      "step": 622
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3e-05,
+      "loss": 0.4312,
+      "step": 623
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3e-05,
+      "loss": 0.5659,
+      "step": 624
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3e-05,
+      "loss": 0.3484,
+      "step": 625
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3e-05,
+      "loss": 0.4551,
+      "step": 626
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3e-05,
+      "loss": 0.314,
+      "step": 627
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3e-05,
+      "loss": 0.3931,
+      "step": 628
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3e-05,
+      "loss": 0.3777,
+      "step": 629
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3e-05,
+      "loss": 0.4124,
+      "step": 630
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3e-05,
+      "loss": 0.4526,
+      "step": 631
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3e-05,
+      "loss": 0.2893,
+      "step": 632
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3e-05,
+      "loss": 0.4595,
+      "step": 633
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3e-05,
+      "loss": 0.4788,
+      "step": 634
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3e-05,
+      "loss": 0.406,
+      "step": 635
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3e-05,
+      "loss": 0.345,
+      "step": 636
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3e-05,
+      "loss": 0.5107,
+      "step": 637
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3e-05,
+      "loss": 0.3477,
+      "step": 638
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3e-05,
+      "loss": 0.3933,
+      "step": 639
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3e-05,
+      "loss": 0.4866,
+      "step": 640
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3e-05,
+      "loss": 0.3254,
+      "step": 641
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3e-05,
+      "loss": 0.2927,
+      "step": 642
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3e-05,
+      "loss": 0.3135,
+      "step": 643
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3e-05,
+      "loss": 0.4346,
+      "step": 644
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3e-05,
+      "loss": 0.5181,
+      "step": 645
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3e-05,
+      "loss": 0.5347,
+      "step": 646
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3e-05,
+      "loss": 0.4775,
+      "step": 647
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3e-05,
+      "loss": 0.3933,
+      "step": 648
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3e-05,
+      "loss": 0.5732,
+      "step": 649
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3e-05,
+      "loss": 0.3474,
+      "step": 650
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3e-05,
+      "loss": 0.3289,
+      "step": 651
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3e-05,
+      "loss": 0.2727,
+      "step": 652
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3e-05,
+      "loss": 0.3059,
+      "step": 653
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3e-05,
+      "loss": 0.5371,
+      "step": 654
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3e-05,
+      "loss": 0.4104,
+      "step": 655
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3e-05,
+      "loss": 0.3357,
+      "step": 656
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3e-05,
+      "loss": 0.3684,
+      "step": 657
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3e-05,
+      "loss": 0.3193,
+      "step": 658
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3e-05,
+      "loss": 0.3096,
+      "step": 659
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3e-05,
+      "loss": 0.5176,
+      "step": 660
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3e-05,
+      "loss": 0.4375,
+      "step": 661
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3e-05,
+      "loss": 0.3674,
+      "step": 662
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3e-05,
+      "loss": 0.426,
+      "step": 663
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3e-05,
+      "loss": 0.3286,
+      "step": 664
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3e-05,
+      "loss": 0.4451,
+      "step": 665
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3e-05,
+      "loss": 0.2917,
+      "step": 666
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3e-05,
+      "loss": 0.4126,
+      "step": 667
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3e-05,
+      "loss": 0.3562,
+      "step": 668
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3e-05,
+      "loss": 0.3228,
+      "step": 669
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3e-05,
+      "loss": 0.2405,
+      "step": 670
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3e-05,
+      "loss": 0.3315,
+      "step": 671
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3e-05,
+      "loss": 0.2517,
+      "step": 672
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3e-05,
+      "loss": 0.4868,
+      "step": 673
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3e-05,
+      "loss": 0.3889,
+      "step": 674
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3e-05,
+      "loss": 0.3469,
+      "step": 675
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3e-05,
+      "loss": 0.4546,
+      "step": 676
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3e-05,
+      "loss": 0.2449,
+      "step": 677
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3e-05,
+      "loss": 0.3826,
+      "step": 678
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3e-05,
+      "loss": 0.3257,
+      "step": 679
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3e-05,
+      "loss": 0.3376,
+      "step": 680
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3e-05,
+      "loss": 0.3772,
+      "step": 681
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3e-05,
+      "loss": 0.3225,
+      "step": 682
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3e-05,
+      "loss": 0.4099,
+      "step": 683
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3e-05,
+      "loss": 0.4758,
+      "step": 684
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3e-05,
+      "loss": 0.3882,
+      "step": 685
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3e-05,
+      "loss": 0.3677,
+      "step": 686
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3e-05,
+      "loss": 0.458,
+      "step": 687
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3e-05,
+      "loss": 0.3909,
+      "step": 688
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 3e-05,
+      "loss": 0.449,
+      "step": 689
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3e-05,
+      "loss": 0.3801,
+      "step": 690
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3e-05,
+      "loss": 0.4229,
+      "step": 691
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 3e-05,
+      "loss": 0.4094,
+      "step": 692
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3e-05,
+      "loss": 0.4116,
+      "step": 693
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 3e-05,
+      "loss": 0.5342,
+      "step": 694
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3e-05,
+      "loss": 0.3535,
+      "step": 695
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3e-05,
+      "loss": 0.283,
+      "step": 696
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 3e-05,
+      "loss": 0.4346,
+      "step": 697
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 3e-05,
+      "loss": 0.356,
+      "step": 698
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3e-05,
+      "loss": 0.4661,
+      "step": 699
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 3e-05,
+      "loss": 0.4143,
+      "step": 700
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3e-05,
+      "loss": 0.4778,
+      "step": 701
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3e-05,
+      "loss": 0.5815,
+      "step": 702
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 3e-05,
+      "loss": 0.459,
+      "step": 703
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3e-05,
+      "loss": 0.3394,
+      "step": 704
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 3e-05,
+      "loss": 0.4946,
+      "step": 705
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3e-05,
+      "loss": 0.3389,
+      "step": 706
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 3e-05,
+      "loss": 0.4587,
+      "step": 707
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 3e-05,
+      "loss": 0.2769,
+      "step": 708
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 3e-05,
+      "loss": 0.3755,
+      "step": 709
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 3e-05,
+      "loss": 0.4326,
+      "step": 710
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3e-05,
+      "loss": 0.4553,
+      "step": 711
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3e-05,
+      "loss": 0.3311,
+      "step": 712
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3e-05,
+      "loss": 0.5107,
+      "step": 713
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3e-05,
+      "loss": 0.521,
+      "step": 714
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3e-05,
+      "loss": 0.6138,
+      "step": 715
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 3e-05,
+      "loss": 0.3347,
+      "step": 716
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 3e-05,
+      "loss": 0.3696,
+      "step": 717
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3e-05,
+      "loss": 0.2771,
+      "step": 718
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3e-05,
+      "loss": 0.428,
+      "step": 719
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3e-05,
+      "loss": 0.2478,
+      "step": 720
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 3e-05,
+      "loss": 0.2729,
+      "step": 721
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3e-05,
+      "loss": 0.4375,
+      "step": 722
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3e-05,
+      "loss": 0.342,
+      "step": 723
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3e-05,
+      "loss": 0.3005,
+      "step": 724
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3e-05,
+      "loss": 0.3474,
+      "step": 725
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3e-05,
+      "loss": 0.3608,
+      "step": 726
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3e-05,
+      "loss": 0.4558,
+      "step": 727
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3e-05,
+      "loss": 0.301,
+      "step": 728
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3e-05,
+      "loss": 0.3081,
+      "step": 729
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3e-05,
+      "loss": 0.3252,
+      "step": 730
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3e-05,
+      "loss": 0.4917,
+      "step": 731
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3e-05,
+      "loss": 0.6221,
+      "step": 732
+    },
+    {
+      "epoch": 6.0,
+      "eval_loss": 0.404296875,
+      "eval_mae": 0.48583984375,
+      "eval_mse": 0.404296875,
+      "eval_runtime": 51.0961,
+      "eval_samples_per_second": 3236.903,
+      "eval_steps_per_second": 0.274,
+      "step": 732
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3e-05,
+      "loss": 0.3599,
+      "step": 733
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3e-05,
+      "loss": 0.3186,
+      "step": 734
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3e-05,
+      "loss": 0.397,
+      "step": 735
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3e-05,
+      "loss": 0.3984,
+      "step": 736
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3e-05,
+      "loss": 0.3057,
+      "step": 737
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3e-05,
+      "loss": 0.3975,
+      "step": 738
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3e-05,
+      "loss": 0.4192,
+      "step": 739
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3e-05,
+      "loss": 0.3003,
+      "step": 740
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3e-05,
+      "loss": 0.2715,
+      "step": 741
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3e-05,
+      "loss": 0.3076,
+      "step": 742
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3e-05,
+      "loss": 0.5244,
+      "step": 743
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3e-05,
+      "loss": 0.3604,
+      "step": 744
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3e-05,
+      "loss": 0.4128,
+      "step": 745
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3e-05,
+      "loss": 0.4688,
+      "step": 746
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3e-05,
+      "loss": 0.3489,
+      "step": 747
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3e-05,
+      "loss": 0.4355,
+      "step": 748
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3e-05,
+      "loss": 0.2847,
+      "step": 749
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3e-05,
+      "loss": 0.304,
+      "step": 750
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3e-05,
+      "loss": 0.3767,
+      "step": 751
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3e-05,
+      "loss": 0.5469,
+      "step": 752
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3e-05,
+      "loss": 0.3091,
+      "step": 753
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3e-05,
+      "loss": 0.3806,
+      "step": 754
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3e-05,
+      "loss": 0.2861,
+      "step": 755
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3e-05,
+      "loss": 0.4626,
+      "step": 756
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3e-05,
+      "loss": 0.3892,
+      "step": 757
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3e-05,
+      "loss": 0.3481,
+      "step": 758
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3e-05,
+      "loss": 0.3928,
+      "step": 759
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3e-05,
+      "loss": 0.5557,
+      "step": 760
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3e-05,
+      "loss": 0.4065,
+      "step": 761
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3e-05,
+      "loss": 0.342,
+      "step": 762
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3e-05,
+      "loss": 0.3357,
+      "step": 763
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3e-05,
+      "loss": 0.4067,
+      "step": 764
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3e-05,
+      "loss": 0.3389,
+      "step": 765
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3e-05,
+      "loss": 0.4336,
+      "step": 766
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3e-05,
+      "loss": 0.2764,
+      "step": 767
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3e-05,
+      "loss": 0.386,
+      "step": 768
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3e-05,
+      "loss": 0.332,
+      "step": 769
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3e-05,
+      "loss": 0.3523,
+      "step": 770
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3e-05,
+      "loss": 0.3687,
+      "step": 771
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3e-05,
+      "loss": 0.3477,
+      "step": 772
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3e-05,
+      "loss": 0.543,
+      "step": 773
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3e-05,
+      "loss": 0.3245,
+      "step": 774
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3e-05,
+      "loss": 0.4172,
+      "step": 775
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3e-05,
+      "loss": 0.219,
+      "step": 776
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3e-05,
+      "loss": 0.4634,
+      "step": 777
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3e-05,
+      "loss": 0.3958,
+      "step": 778
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3e-05,
+      "loss": 0.3828,
+      "step": 779
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3e-05,
+      "loss": 0.3596,
+      "step": 780
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3e-05,
+      "loss": 0.6064,
+      "step": 781
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3e-05,
+      "loss": 0.3132,
+      "step": 782
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3e-05,
+      "loss": 0.3069,
+      "step": 783
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3e-05,
+      "loss": 0.3071,
+      "step": 784
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3e-05,
+      "loss": 0.3064,
+      "step": 785
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3e-05,
+      "loss": 0.3943,
+      "step": 786
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3e-05,
+      "loss": 0.3569,
+      "step": 787
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3e-05,
+      "loss": 0.3354,
+      "step": 788
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3e-05,
+      "loss": 0.3225,
+      "step": 789
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3e-05,
+      "loss": 0.3354,
+      "step": 790
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3e-05,
+      "loss": 0.2805,
+      "step": 791
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3e-05,
+      "loss": 0.4155,
+      "step": 792
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3e-05,
+      "loss": 0.4497,
+      "step": 793
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3e-05,
+      "loss": 0.3972,
+      "step": 794
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3e-05,
+      "loss": 0.3057,
+      "step": 795
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3e-05,
+      "loss": 0.5664,
+      "step": 796
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3e-05,
+      "loss": 0.5239,
+      "step": 797
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3e-05,
+      "loss": 0.625,
+      "step": 798
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3e-05,
+      "loss": 0.3101,
+      "step": 799
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3e-05,
+      "loss": 0.3279,
+      "step": 800
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3e-05,
+      "loss": 0.4138,
+      "step": 801
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3e-05,
+      "loss": 0.3069,
+      "step": 802
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3e-05,
+      "loss": 0.3179,
+      "step": 803
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3e-05,
+      "loss": 0.3022,
+      "step": 804
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3e-05,
+      "loss": 0.3481,
+      "step": 805
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3e-05,
+      "loss": 0.2607,
+      "step": 806
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3e-05,
+      "loss": 0.4023,
+      "step": 807
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3e-05,
+      "loss": 0.3481,
+      "step": 808
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3e-05,
+      "loss": 0.5376,
+      "step": 809
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3e-05,
+      "loss": 0.3123,
+      "step": 810
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3e-05,
+      "loss": 0.457,
+      "step": 811
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3e-05,
+      "loss": 0.4941,
+      "step": 812
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3e-05,
+      "loss": 0.4426,
+      "step": 813
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3e-05,
+      "loss": 0.3079,
+      "step": 814
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3e-05,
+      "loss": 0.2891,
+      "step": 815
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3e-05,
+      "loss": 0.3562,
+      "step": 816
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3e-05,
+      "loss": 0.4565,
+      "step": 817
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3e-05,
+      "loss": 0.2983,
+      "step": 818
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3e-05,
+      "loss": 0.3252,
+      "step": 819
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3e-05,
+      "loss": 0.3984,
+      "step": 820
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3e-05,
+      "loss": 0.3745,
+      "step": 821
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3e-05,
+      "loss": 0.4111,
+      "step": 822
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3e-05,
+      "loss": 0.5991,
+      "step": 823
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3e-05,
+      "loss": 0.3647,
+      "step": 824
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3e-05,
+      "loss": 0.334,
+      "step": 825
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3e-05,
+      "loss": 0.4312,
+      "step": 826
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3e-05,
+      "loss": 0.4932,
+      "step": 827
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3e-05,
+      "loss": 0.3181,
+      "step": 828
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3e-05,
+      "loss": 0.4404,
+      "step": 829
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3e-05,
+      "loss": 0.3662,
+      "step": 830
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3e-05,
+      "loss": 0.324,
+      "step": 831
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3e-05,
+      "loss": 0.2666,
+      "step": 832
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3e-05,
+      "loss": 0.4375,
+      "step": 833
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3e-05,
+      "loss": 0.3599,
+      "step": 834
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3e-05,
+      "loss": 0.3132,
+      "step": 835
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3e-05,
+      "loss": 0.3254,
+      "step": 836
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3e-05,
+      "loss": 0.2334,
+      "step": 837
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3e-05,
+      "loss": 0.2666,
+      "step": 838
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3e-05,
+      "loss": 0.3723,
+      "step": 839
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3e-05,
+      "loss": 0.3438,
+      "step": 840
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3e-05,
+      "loss": 0.3755,
+      "step": 841
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3e-05,
+      "loss": 0.3372,
+      "step": 842
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3e-05,
+      "loss": 0.3428,
+      "step": 843
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3e-05,
+      "loss": 0.2556,
+      "step": 844
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3e-05,
+      "loss": 0.4324,
+      "step": 845
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3e-05,
+      "loss": 0.5083,
+      "step": 846
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3e-05,
+      "loss": 0.6929,
+      "step": 847
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3e-05,
+      "loss": 0.3386,
+      "step": 848
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3e-05,
+      "loss": 0.4075,
+      "step": 849
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3e-05,
+      "loss": 0.4507,
+      "step": 850
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3e-05,
+      "loss": 0.3223,
+      "step": 851
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3e-05,
+      "loss": 0.3071,
+      "step": 852
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3e-05,
+      "loss": 0.4441,
+      "step": 853
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3e-05,
+      "loss": 0.2295,
+      "step": 854
+    },
+    {
+      "epoch": 7.0,
+      "eval_loss": 0.38232421875,
+      "eval_mae": 0.47119140625,
+      "eval_mse": 0.38232421875,
+      "eval_runtime": 51.1323,
+      "eval_samples_per_second": 3234.608,
+      "eval_steps_per_second": 0.274,
+      "step": 854
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3e-05,
+      "loss": 0.4712,
+      "step": 855
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3e-05,
+      "loss": 0.3088,
+      "step": 856
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3e-05,
+      "loss": 0.29,
+      "step": 857
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3e-05,
+      "loss": 0.4451,
+      "step": 858
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3e-05,
+      "loss": 0.2502,
+      "step": 859
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3e-05,
+      "loss": 0.3379,
+      "step": 860
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3e-05,
+      "loss": 0.3606,
+      "step": 861
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3e-05,
+      "loss": 0.4585,
+      "step": 862
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3e-05,
+      "loss": 0.3784,
+      "step": 863
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3e-05,
+      "loss": 0.408,
+      "step": 864
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3e-05,
+      "loss": 0.2783,
+      "step": 865
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3e-05,
+      "loss": 0.314,
+      "step": 866
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3e-05,
+      "loss": 0.3159,
+      "step": 867
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3e-05,
+      "loss": 0.3774,
+      "step": 868
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3e-05,
+      "loss": 0.2993,
+      "step": 869
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3e-05,
+      "loss": 0.4478,
+      "step": 870
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3e-05,
+      "loss": 0.3748,
+      "step": 871
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3e-05,
+      "loss": 0.4021,
+      "step": 872
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3e-05,
+      "loss": 0.2408,
+      "step": 873
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3e-05,
+      "loss": 0.3887,
+      "step": 874
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3e-05,
+      "loss": 0.363,
+      "step": 875
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3e-05,
+      "loss": 0.4795,
+      "step": 876
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3e-05,
+      "loss": 0.4031,
+      "step": 877
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3e-05,
+      "loss": 0.5063,
+      "step": 878
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3e-05,
+      "loss": 0.366,
+      "step": 879
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3e-05,
+      "loss": 0.2698,
+      "step": 880
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3e-05,
+      "loss": 0.3169,
+      "step": 881
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3e-05,
+      "loss": 0.373,
+      "step": 882
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3e-05,
+      "loss": 0.3076,
+      "step": 883
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3e-05,
+      "loss": 0.4355,
+      "step": 884
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3e-05,
+      "loss": 0.4648,
+      "step": 885
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3e-05,
+      "loss": 0.343,
+      "step": 886
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3e-05,
+      "loss": 0.3267,
+      "step": 887
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3e-05,
+      "loss": 0.3123,
+      "step": 888
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3e-05,
+      "loss": 0.3342,
+      "step": 889
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3e-05,
+      "loss": 0.406,
+      "step": 890
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3e-05,
+      "loss": 0.2578,
+      "step": 891
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3e-05,
+      "loss": 0.3452,
+      "step": 892
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3e-05,
+      "loss": 0.428,
+      "step": 893
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3e-05,
+      "loss": 0.373,
+      "step": 894
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3e-05,
+      "loss": 0.3816,
+      "step": 895
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3e-05,
+      "loss": 0.271,
+      "step": 896
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3e-05,
+      "loss": 0.5537,
+      "step": 897
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3e-05,
+      "loss": 0.375,
+      "step": 898
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3e-05,
+      "loss": 0.3398,
+      "step": 899
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3e-05,
+      "loss": 0.4387,
+      "step": 900
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3e-05,
+      "loss": 0.2766,
+      "step": 901
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3e-05,
+      "loss": 0.4011,
+      "step": 902
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3e-05,
+      "loss": 0.3186,
+      "step": 903
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3e-05,
+      "loss": 0.282,
+      "step": 904
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3e-05,
+      "loss": 0.4431,
+      "step": 905
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3e-05,
+      "loss": 0.4565,
+      "step": 906
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3e-05,
+      "loss": 0.3162,
+      "step": 907
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3e-05,
+      "loss": 0.2859,
+      "step": 908
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3e-05,
+      "loss": 0.3313,
+      "step": 909
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3e-05,
+      "loss": 0.3906,
+      "step": 910
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3e-05,
+      "loss": 0.2686,
+      "step": 911
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3e-05,
+      "loss": 0.4556,
+      "step": 912
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3e-05,
+      "loss": 0.3899,
+      "step": 913
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3e-05,
+      "loss": 0.3625,
+      "step": 914
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3e-05,
+      "loss": 0.4011,
+      "step": 915
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3e-05,
+      "loss": 0.4285,
+      "step": 916
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3e-05,
+      "loss": 0.2664,
+      "step": 917
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3e-05,
+      "loss": 0.5132,
+      "step": 918
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3e-05,
+      "loss": 0.4363,
+      "step": 919
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3e-05,
+      "loss": 0.2986,
+      "step": 920
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3e-05,
+      "loss": 0.3196,
+      "step": 921
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3e-05,
+      "loss": 0.3345,
+      "step": 922
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3e-05,
+      "loss": 0.4143,
+      "step": 923
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3e-05,
+      "loss": 0.4443,
+      "step": 924
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3e-05,
+      "loss": 0.2739,
+      "step": 925
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3e-05,
+      "loss": 0.3911,
+      "step": 926
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3e-05,
+      "loss": 0.3203,
+      "step": 927
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3e-05,
+      "loss": 0.2118,
+      "step": 928
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3e-05,
+      "loss": 0.2993,
+      "step": 929
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3e-05,
+      "loss": 0.3381,
+      "step": 930
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3e-05,
+      "loss": 0.4219,
+      "step": 931
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3e-05,
+      "loss": 0.4807,
+      "step": 932
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3e-05,
+      "loss": 0.5186,
+      "step": 933
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3e-05,
+      "loss": 0.468,
+      "step": 934
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3e-05,
+      "loss": 0.2271,
+      "step": 935
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3e-05,
+      "loss": 0.4446,
+      "step": 936
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3e-05,
+      "loss": 0.4121,
+      "step": 937
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3e-05,
+      "loss": 0.489,
+      "step": 938
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3e-05,
+      "loss": 0.4153,
+      "step": 939
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3e-05,
+      "loss": 0.6919,
+      "step": 940
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3e-05,
+      "loss": 0.2793,
+      "step": 941
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3e-05,
+      "loss": 0.3538,
+      "step": 942
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3e-05,
+      "loss": 0.2617,
+      "step": 943
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3e-05,
+      "loss": 0.3582,
+      "step": 944
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3e-05,
+      "loss": 0.3521,
+      "step": 945
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3e-05,
+      "loss": 0.397,
+      "step": 946
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3e-05,
+      "loss": 0.3015,
+      "step": 947
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3e-05,
+      "loss": 0.2515,
+      "step": 948
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3e-05,
+      "loss": 0.3904,
+      "step": 949
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3e-05,
+      "loss": 0.303,
+      "step": 950
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3e-05,
+      "loss": 0.3479,
+      "step": 951
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3e-05,
+      "loss": 0.3303,
+      "step": 952
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3e-05,
+      "loss": 0.4333,
+      "step": 953
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3e-05,
+      "loss": 0.3518,
+      "step": 954
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3e-05,
+      "loss": 0.2362,
+      "step": 955
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3e-05,
+      "loss": 0.4619,
+      "step": 956
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3e-05,
+      "loss": 0.3784,
+      "step": 957
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3e-05,
+      "loss": 0.2759,
+      "step": 958
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3e-05,
+      "loss": 0.4866,
+      "step": 959
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3e-05,
+      "loss": 0.3062,
+      "step": 960
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3e-05,
+      "loss": 0.4597,
+      "step": 961
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3e-05,
+      "loss": 0.3481,
+      "step": 962
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3e-05,
+      "loss": 0.4065,
+      "step": 963
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3e-05,
+      "loss": 0.2991,
+      "step": 964
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3e-05,
+      "loss": 0.2649,
+      "step": 965
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3e-05,
+      "loss": 0.4287,
+      "step": 966
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3e-05,
+      "loss": 0.3975,
+      "step": 967
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3e-05,
+      "loss": 0.2922,
+      "step": 968
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3e-05,
+      "loss": 0.3967,
+      "step": 969
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3e-05,
+      "loss": 0.4412,
+      "step": 970
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3e-05,
+      "loss": 0.3337,
+      "step": 971
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3e-05,
+      "loss": 0.3633,
+      "step": 972
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3e-05,
+      "loss": 0.6152,
+      "step": 973
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3e-05,
+      "loss": 0.3008,
+      "step": 974
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3e-05,
+      "loss": 0.2384,
+      "step": 975
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3e-05,
+      "loss": 0.6694,
+      "step": 976
+    },
+    {
+      "epoch": 8.0,
+      "eval_loss": 0.37109375,
+      "eval_mae": 0.46240234375,
+      "eval_mse": 0.37109375,
+      "eval_runtime": 51.1469,
+      "eval_samples_per_second": 3233.687,
+      "eval_steps_per_second": 0.274,
+      "step": 976
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 3e-05,
+      "loss": 0.2864,
+      "step": 977
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3e-05,
+      "loss": 0.2957,
+      "step": 978
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3e-05,
+      "loss": 0.3149,
+      "step": 979
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3e-05,
+      "loss": 0.3826,
+      "step": 980
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3e-05,
+      "loss": 0.3643,
+      "step": 981
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 3e-05,
+      "loss": 0.4324,
+      "step": 982
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3e-05,
+      "loss": 0.3191,
+      "step": 983
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3e-05,
+      "loss": 0.3008,
+      "step": 984
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3e-05,
+      "loss": 0.2842,
+      "step": 985
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 3e-05,
+      "loss": 0.6108,
+      "step": 986
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3e-05,
+      "loss": 0.3318,
+      "step": 987
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3e-05,
+      "loss": 0.3955,
+      "step": 988
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3e-05,
+      "loss": 0.396,
+      "step": 989
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3e-05,
+      "loss": 0.2825,
+      "step": 990
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 3e-05,
+      "loss": 0.2491,
+      "step": 991
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3e-05,
+      "loss": 0.5303,
+      "step": 992
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3e-05,
+      "loss": 0.6816,
+      "step": 993
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 3e-05,
+      "loss": 0.3894,
+      "step": 994
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1989,
+      "step": 995
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 3e-05,
+      "loss": 0.3669,
+      "step": 996
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3e-05,
+      "loss": 0.4126,
+      "step": 997
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 3e-05,
+      "loss": 0.4509,
+      "step": 998
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 3e-05,
+      "loss": 0.2418,
+      "step": 999
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3e-05,
+      "loss": 0.2271,
+      "step": 1000
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3e-05,
+      "loss": 0.3203,
+      "step": 1001
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3e-05,
+      "loss": 0.4192,
+      "step": 1002
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3e-05,
+      "loss": 0.3623,
+      "step": 1003
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 3e-05,
+      "loss": 0.3464,
+      "step": 1004
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3e-05,
+      "loss": 0.3132,
+      "step": 1005
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1721,
+      "step": 1006
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 3e-05,
+      "loss": 0.2864,
+      "step": 1007
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3e-05,
+      "loss": 0.5024,
+      "step": 1008
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3e-05,
+      "loss": 0.3237,
+      "step": 1009
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 3e-05,
+      "loss": 0.3579,
+      "step": 1010
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3e-05,
+      "loss": 0.3474,
+      "step": 1011
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1741,
+      "step": 1012
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 3e-05,
+      "loss": 0.3496,
+      "step": 1013
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3e-05,
+      "loss": 0.3359,
+      "step": 1014
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 3e-05,
+      "loss": 0.3923,
+      "step": 1015
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 3e-05,
+      "loss": 0.2695,
+      "step": 1016
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3e-05,
+      "loss": 0.3726,
+      "step": 1017
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3e-05,
+      "loss": 0.2971,
+      "step": 1018
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3e-05,
+      "loss": 0.2766,
+      "step": 1019
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 3e-05,
+      "loss": 0.2316,
+      "step": 1020
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 3e-05,
+      "loss": 0.5205,
+      "step": 1021
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3e-05,
+      "loss": 0.3489,
+      "step": 1022
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3e-05,
+      "loss": 0.4175,
+      "step": 1023
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 3e-05,
+      "loss": 0.22,
+      "step": 1024
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3e-05,
+      "loss": 0.2852,
+      "step": 1025
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 3e-05,
+      "loss": 0.2976,
+      "step": 1026
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3e-05,
+      "loss": 0.3904,
+      "step": 1027
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3e-05,
+      "loss": 0.4705,
+      "step": 1028
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 3e-05,
+      "loss": 0.2393,
+      "step": 1029
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3e-05,
+      "loss": 0.2529,
+      "step": 1030
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 3e-05,
+      "loss": 0.29,
+      "step": 1031
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 3e-05,
+      "loss": 0.2423,
+      "step": 1032
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 3e-05,
+      "loss": 0.415,
+      "step": 1033
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3e-05,
+      "loss": 0.4033,
+      "step": 1034
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3e-05,
+      "loss": 0.2815,
+      "step": 1035
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3e-05,
+      "loss": 0.2744,
+      "step": 1036
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 3e-05,
+      "loss": 0.4561,
+      "step": 1037
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3e-05,
+      "loss": 0.3411,
+      "step": 1038
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3e-05,
+      "loss": 0.5078,
+      "step": 1039
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 3e-05,
+      "loss": 0.4443,
+      "step": 1040
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3e-05,
+      "loss": 0.4265,
+      "step": 1041
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 3e-05,
+      "loss": 0.3494,
+      "step": 1042
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3e-05,
+      "loss": 0.3401,
+      "step": 1043
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 3e-05,
+      "loss": 0.4177,
+      "step": 1044
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3e-05,
+      "loss": 0.3101,
+      "step": 1045
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3e-05,
+      "loss": 0.251,
+      "step": 1046
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3e-05,
+      "loss": 0.2349,
+      "step": 1047
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 3e-05,
+      "loss": 0.2952,
+      "step": 1048
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3e-05,
+      "loss": 0.2603,
+      "step": 1049
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3e-05,
+      "loss": 0.415,
+      "step": 1050
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 3e-05,
+      "loss": 0.3274,
+      "step": 1051
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3e-05,
+      "loss": 0.3318,
+      "step": 1052
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 3e-05,
+      "loss": 0.3218,
+      "step": 1053
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3e-05,
+      "loss": 0.2292,
+      "step": 1054
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3e-05,
+      "loss": 0.4985,
+      "step": 1055
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3e-05,
+      "loss": 0.3757,
+      "step": 1056
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3e-05,
+      "loss": 0.3379,
+      "step": 1057
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3e-05,
+      "loss": 0.3264,
+      "step": 1058
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3e-05,
+      "loss": 0.4634,
+      "step": 1059
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3e-05,
+      "loss": 0.2883,
+      "step": 1060
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3e-05,
+      "loss": 0.2983,
+      "step": 1061
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3e-05,
+      "loss": 0.4292,
+      "step": 1062
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3e-05,
+      "loss": 0.3716,
+      "step": 1063
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3e-05,
+      "loss": 0.2632,
+      "step": 1064
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3e-05,
+      "loss": 0.4045,
+      "step": 1065
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3e-05,
+      "loss": 0.3557,
+      "step": 1066
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3e-05,
+      "loss": 0.2822,
+      "step": 1067
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3e-05,
+      "loss": 0.2373,
+      "step": 1068
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3e-05,
+      "loss": 0.4377,
+      "step": 1069
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3e-05,
+      "loss": 0.4177,
+      "step": 1070
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3e-05,
+      "loss": 0.2874,
+      "step": 1071
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3e-05,
+      "loss": 0.2241,
+      "step": 1072
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3e-05,
+      "loss": 0.3411,
+      "step": 1073
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3e-05,
+      "loss": 0.2512,
+      "step": 1074
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3e-05,
+      "loss": 0.3286,
+      "step": 1075
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3e-05,
+      "loss": 0.3604,
+      "step": 1076
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3e-05,
+      "loss": 0.3752,
+      "step": 1077
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3e-05,
+      "loss": 0.3555,
+      "step": 1078
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3e-05,
+      "loss": 0.3127,
+      "step": 1079
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3e-05,
+      "loss": 0.2996,
+      "step": 1080
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3e-05,
+      "loss": 0.3074,
+      "step": 1081
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3e-05,
+      "loss": 0.3462,
+      "step": 1082
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3e-05,
+      "loss": 0.5742,
+      "step": 1083
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3e-05,
+      "loss": 0.3687,
+      "step": 1084
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3e-05,
+      "loss": 0.2996,
+      "step": 1085
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3e-05,
+      "loss": 0.2671,
+      "step": 1086
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3e-05,
+      "loss": 0.4556,
+      "step": 1087
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3e-05,
+      "loss": 0.3201,
+      "step": 1088
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3e-05,
+      "loss": 0.2437,
+      "step": 1089
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3e-05,
+      "loss": 0.4124,
+      "step": 1090
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3e-05,
+      "loss": 0.3301,
+      "step": 1091
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3e-05,
+      "loss": 0.2739,
+      "step": 1092
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3e-05,
+      "loss": 0.4065,
+      "step": 1093
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3e-05,
+      "loss": 0.2744,
+      "step": 1094
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3e-05,
+      "loss": 0.2725,
+      "step": 1095
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3e-05,
+      "loss": 0.2455,
+      "step": 1096
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3e-05,
+      "loss": 0.2825,
+      "step": 1097
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3e-05,
+      "loss": 0.3164,
+      "step": 1098
+    },
+    {
+      "epoch": 9.0,
+      "eval_loss": 0.353515625,
+      "eval_mae": 0.450439453125,
+      "eval_mse": 0.353515625,
+      "eval_runtime": 51.1182,
+      "eval_samples_per_second": 3235.504,
+      "eval_steps_per_second": 0.274,
+      "step": 1098
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3e-05,
+      "loss": 0.342,
+      "step": 1099
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3e-05,
+      "loss": 0.3665,
+      "step": 1100
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3e-05,
+      "loss": 0.3721,
+      "step": 1101
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3e-05,
+      "loss": 0.243,
+      "step": 1102
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3e-05,
+      "loss": 0.2832,
+      "step": 1103
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3e-05,
+      "loss": 0.3254,
+      "step": 1104
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3e-05,
+      "loss": 0.3315,
+      "step": 1105
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3e-05,
+      "loss": 0.4458,
+      "step": 1106
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3e-05,
+      "loss": 0.3955,
+      "step": 1107
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3e-05,
+      "loss": 0.3296,
+      "step": 1108
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3e-05,
+      "loss": 0.2605,
+      "step": 1109
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3e-05,
+      "loss": 0.2966,
+      "step": 1110
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3e-05,
+      "loss": 0.27,
+      "step": 1111
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3e-05,
+      "loss": 0.3164,
+      "step": 1112
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3e-05,
+      "loss": 0.3735,
+      "step": 1113
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3e-05,
+      "loss": 0.2756,
+      "step": 1114
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3e-05,
+      "loss": 0.3352,
+      "step": 1115
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3e-05,
+      "loss": 0.3879,
+      "step": 1116
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3e-05,
+      "loss": 0.2649,
+      "step": 1117
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3e-05,
+      "loss": 0.3499,
+      "step": 1118
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3e-05,
+      "loss": 0.3035,
+      "step": 1119
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3e-05,
+      "loss": 0.3157,
+      "step": 1120
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3e-05,
+      "loss": 0.3469,
+      "step": 1121
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3e-05,
+      "loss": 0.2932,
+      "step": 1122
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3e-05,
+      "loss": 0.2742,
+      "step": 1123
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3e-05,
+      "loss": 0.2881,
+      "step": 1124
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3e-05,
+      "loss": 0.2939,
+      "step": 1125
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3e-05,
+      "loss": 0.3638,
+      "step": 1126
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3e-05,
+      "loss": 0.2703,
+      "step": 1127
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3e-05,
+      "loss": 0.3232,
+      "step": 1128
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3e-05,
+      "loss": 0.2913,
+      "step": 1129
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3e-05,
+      "loss": 0.3298,
+      "step": 1130
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3e-05,
+      "loss": 0.4297,
+      "step": 1131
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3e-05,
+      "loss": 0.3252,
+      "step": 1132
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3e-05,
+      "loss": 0.2573,
+      "step": 1133
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3e-05,
+      "loss": 0.3684,
+      "step": 1134
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3e-05,
+      "loss": 0.3599,
+      "step": 1135
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3e-05,
+      "loss": 0.2974,
+      "step": 1136
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3e-05,
+      "loss": 0.1869,
+      "step": 1137
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3e-05,
+      "loss": 0.3237,
+      "step": 1138
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3e-05,
+      "loss": 0.304,
+      "step": 1139
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3e-05,
+      "loss": 0.3872,
+      "step": 1140
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3e-05,
+      "loss": 0.2903,
+      "step": 1141
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3e-05,
+      "loss": 0.1965,
+      "step": 1142
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3e-05,
+      "loss": 0.3477,
+      "step": 1143
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3e-05,
+      "loss": 0.3459,
+      "step": 1144
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3e-05,
+      "loss": 0.2458,
+      "step": 1145
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3e-05,
+      "loss": 0.3601,
+      "step": 1146
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3e-05,
+      "loss": 0.323,
+      "step": 1147
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3e-05,
+      "loss": 0.4338,
+      "step": 1148
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 3e-05,
+      "loss": 0.3032,
+      "step": 1149
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3e-05,
+      "loss": 0.4636,
+      "step": 1150
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3e-05,
+      "loss": 0.2549,
+      "step": 1151
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3e-05,
+      "loss": 0.3035,
+      "step": 1152
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 3e-05,
+      "loss": 0.387,
+      "step": 1153
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 3e-05,
+      "loss": 0.377,
+      "step": 1154
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3e-05,
+      "loss": 0.4163,
+      "step": 1155
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 3e-05,
+      "loss": 0.2754,
+      "step": 1156
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 3e-05,
+      "loss": 0.324,
+      "step": 1157
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3e-05,
+      "loss": 0.3167,
+      "step": 1158
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 3e-05,
+      "loss": 0.3298,
+      "step": 1159
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3e-05,
+      "loss": 0.3911,
+      "step": 1160
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3e-05,
+      "loss": 0.4763,
+      "step": 1161
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3e-05,
+      "loss": 0.394,
+      "step": 1162
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 3e-05,
+      "loss": 0.1857,
+      "step": 1163
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3e-05,
+      "loss": 0.2939,
+      "step": 1164
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 3e-05,
+      "loss": 0.2983,
+      "step": 1165
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 3e-05,
+      "loss": 0.2375,
+      "step": 1166
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3e-05,
+      "loss": 0.3811,
+      "step": 1167
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3e-05,
+      "loss": 0.2561,
+      "step": 1168
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 3e-05,
+      "loss": 0.5264,
+      "step": 1169
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3e-05,
+      "loss": 0.313,
+      "step": 1170
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3e-05,
+      "loss": 0.3167,
+      "step": 1171
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3e-05,
+      "loss": 0.3542,
+      "step": 1172
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3e-05,
+      "loss": 0.4165,
+      "step": 1173
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3e-05,
+      "loss": 0.3123,
+      "step": 1174
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3e-05,
+      "loss": 0.3372,
+      "step": 1175
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3e-05,
+      "loss": 0.6162,
+      "step": 1176
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3e-05,
+      "loss": 0.3606,
+      "step": 1177
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3e-05,
+      "loss": 0.3611,
+      "step": 1178
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3e-05,
+      "loss": 0.3201,
+      "step": 1179
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3e-05,
+      "loss": 0.3635,
+      "step": 1180
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3e-05,
+      "loss": 0.3391,
+      "step": 1181
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3e-05,
+      "loss": 0.3015,
+      "step": 1182
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3e-05,
+      "loss": 0.2352,
+      "step": 1183
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3e-05,
+      "loss": 0.3323,
+      "step": 1184
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3e-05,
+      "loss": 0.3506,
+      "step": 1185
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3e-05,
+      "loss": 0.426,
+      "step": 1186
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3e-05,
+      "loss": 0.226,
+      "step": 1187
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3e-05,
+      "loss": 0.6396,
+      "step": 1188
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3e-05,
+      "loss": 0.3662,
+      "step": 1189
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3e-05,
+      "loss": 0.2959,
+      "step": 1190
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3e-05,
+      "loss": 0.5298,
+      "step": 1191
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3e-05,
+      "loss": 0.224,
+      "step": 1192
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3e-05,
+      "loss": 0.2896,
+      "step": 1193
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3e-05,
+      "loss": 0.2839,
+      "step": 1194
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 3e-05,
+      "loss": 0.3179,
+      "step": 1195
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 3e-05,
+      "loss": 0.3247,
+      "step": 1196
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3e-05,
+      "loss": 0.333,
+      "step": 1197
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 3e-05,
+      "loss": 0.1923,
+      "step": 1198
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 3e-05,
+      "loss": 0.3669,
+      "step": 1199
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3e-05,
+      "loss": 0.2329,
+      "step": 1200
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3e-05,
+      "loss": 0.3789,
+      "step": 1201
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 3e-05,
+      "loss": 0.3477,
+      "step": 1202
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3e-05,
+      "loss": 0.3447,
+      "step": 1203
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3e-05,
+      "loss": 0.3003,
+      "step": 1204
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3e-05,
+      "loss": 0.4185,
+      "step": 1205
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3e-05,
+      "loss": 0.2625,
+      "step": 1206
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3e-05,
+      "loss": 0.2996,
+      "step": 1207
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3e-05,
+      "loss": 0.3801,
+      "step": 1208
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3e-05,
+      "loss": 0.2549,
+      "step": 1209
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3e-05,
+      "loss": 0.2849,
+      "step": 1210
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3e-05,
+      "loss": 0.3464,
+      "step": 1211
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3e-05,
+      "loss": 0.6392,
+      "step": 1212
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3e-05,
+      "loss": 0.2583,
+      "step": 1213
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3e-05,
+      "loss": 0.2651,
+      "step": 1214
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3e-05,
+      "loss": 0.252,
+      "step": 1215
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3e-05,
+      "loss": 0.1917,
+      "step": 1216
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3e-05,
+      "loss": 0.2786,
+      "step": 1217
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3e-05,
+      "loss": 0.2847,
+      "step": 1218
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3e-05,
+      "loss": 0.261,
+      "step": 1219
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3e-05,
+      "loss": 0.269,
+      "step": 1220
+    },
+    {
+      "epoch": 10.0,
+      "eval_loss": 0.3427734375,
+      "eval_mae": 0.441650390625,
+      "eval_mse": 0.3427734375,
+      "eval_runtime": 51.1443,
+      "eval_samples_per_second": 3233.853,
+      "eval_steps_per_second": 0.274,
+      "step": 1220
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3e-05,
+      "loss": 0.2119,
+      "step": 1221
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 3e-05,
+      "loss": 0.3538,
+      "step": 1222
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 3e-05,
+      "loss": 0.2386,
+      "step": 1223
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 3e-05,
+      "loss": 0.1953,
+      "step": 1224
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3e-05,
+      "loss": 0.236,
+      "step": 1225
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3e-05,
+      "loss": 0.2673,
+      "step": 1226
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 3e-05,
+      "loss": 0.3313,
+      "step": 1227
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 3e-05,
+      "loss": 0.3552,
+      "step": 1228
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 3e-05,
+      "loss": 0.314,
+      "step": 1229
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 3e-05,
+      "loss": 0.293,
+      "step": 1230
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3e-05,
+      "loss": 0.3875,
+      "step": 1231
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 3e-05,
+      "loss": 0.3064,
+      "step": 1232
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 3e-05,
+      "loss": 0.2715,
+      "step": 1233
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 3e-05,
+      "loss": 0.2273,
+      "step": 1234
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 3e-05,
+      "loss": 0.3047,
+      "step": 1235
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 3e-05,
+      "loss": 0.3308,
+      "step": 1236
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 3e-05,
+      "loss": 0.249,
+      "step": 1237
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 3e-05,
+      "loss": 0.3396,
+      "step": 1238
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 3e-05,
+      "loss": 0.2856,
+      "step": 1239
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 3e-05,
+      "loss": 0.2194,
+      "step": 1240
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 3e-05,
+      "loss": 0.3823,
+      "step": 1241
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3e-05,
+      "loss": 0.3167,
+      "step": 1242
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3e-05,
+      "loss": 0.3003,
+      "step": 1243
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3e-05,
+      "loss": 0.3777,
+      "step": 1244
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3e-05,
+      "loss": 0.4026,
+      "step": 1245
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3e-05,
+      "loss": 0.2952,
+      "step": 1246
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3e-05,
+      "loss": 0.2625,
+      "step": 1247
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3e-05,
+      "loss": 0.3589,
+      "step": 1248
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 3e-05,
+      "loss": 0.2661,
+      "step": 1249
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3e-05,
+      "loss": 0.2588,
+      "step": 1250
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3e-05,
+      "loss": 0.2808,
+      "step": 1251
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3e-05,
+      "loss": 0.2678,
+      "step": 1252
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 3e-05,
+      "loss": 0.4092,
+      "step": 1253
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 3e-05,
+      "loss": 0.2549,
+      "step": 1254
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 3e-05,
+      "loss": 0.2443,
+      "step": 1255
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 3e-05,
+      "loss": 0.3462,
+      "step": 1256
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 3e-05,
+      "loss": 0.2864,
+      "step": 1257
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 3e-05,
+      "loss": 0.386,
+      "step": 1258
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 3e-05,
+      "loss": 0.2703,
+      "step": 1259
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 3e-05,
+      "loss": 0.2566,
+      "step": 1260
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 3e-05,
+      "loss": 0.2737,
+      "step": 1261
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 3e-05,
+      "loss": 0.4653,
+      "step": 1262
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 3e-05,
+      "loss": 0.2499,
+      "step": 1263
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 3e-05,
+      "loss": 0.2703,
+      "step": 1264
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 3e-05,
+      "loss": 0.3245,
+      "step": 1265
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 3e-05,
+      "loss": 0.3201,
+      "step": 1266
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 3e-05,
+      "loss": 0.2546,
+      "step": 1267
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 3e-05,
+      "loss": 0.2961,
+      "step": 1268
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 3e-05,
+      "loss": 0.3245,
+      "step": 1269
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 3e-05,
+      "loss": 0.3147,
+      "step": 1270
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 3e-05,
+      "loss": 0.261,
+      "step": 1271
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 3e-05,
+      "loss": 0.215,
+      "step": 1272
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 3e-05,
+      "loss": 0.203,
+      "step": 1273
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 3e-05,
+      "loss": 0.3723,
+      "step": 1274
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 3e-05,
+      "loss": 0.4304,
+      "step": 1275
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 3e-05,
+      "loss": 0.4844,
+      "step": 1276
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 3e-05,
+      "loss": 0.3098,
+      "step": 1277
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 3e-05,
+      "loss": 0.2306,
+      "step": 1278
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 3e-05,
+      "loss": 0.3594,
+      "step": 1279
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 3e-05,
+      "loss": 0.2388,
+      "step": 1280
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 3e-05,
+      "loss": 0.2871,
+      "step": 1281
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 3e-05,
+      "loss": 0.2664,
+      "step": 1282
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 3e-05,
+      "loss": 0.2175,
+      "step": 1283
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 3e-05,
+      "loss": 0.2256,
+      "step": 1284
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 3e-05,
+      "loss": 0.2776,
+      "step": 1285
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 3e-05,
+      "loss": 0.3115,
+      "step": 1286
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 3e-05,
+      "loss": 0.2177,
+      "step": 1287
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 3e-05,
+      "loss": 0.2896,
+      "step": 1288
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 3e-05,
+      "loss": 0.2485,
+      "step": 1289
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 3e-05,
+      "loss": 0.3132,
+      "step": 1290
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 3e-05,
+      "loss": 0.4346,
+      "step": 1291
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 3e-05,
+      "loss": 0.2661,
+      "step": 1292
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 3e-05,
+      "loss": 0.2952,
+      "step": 1293
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 3e-05,
+      "loss": 0.29,
+      "step": 1294
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 3e-05,
+      "loss": 0.3223,
+      "step": 1295
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 3e-05,
+      "loss": 0.2595,
+      "step": 1296
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 3e-05,
+      "loss": 0.3416,
+      "step": 1297
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 3e-05,
+      "loss": 0.4294,
+      "step": 1298
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 3e-05,
+      "loss": 0.3674,
+      "step": 1299
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 3e-05,
+      "loss": 0.3904,
+      "step": 1300
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 3e-05,
+      "loss": 0.2681,
+      "step": 1301
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 3e-05,
+      "loss": 0.3032,
+      "step": 1302
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 3e-05,
+      "loss": 0.4824,
+      "step": 1303
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 3e-05,
+      "loss": 0.2671,
+      "step": 1304
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 3e-05,
+      "loss": 0.4194,
+      "step": 1305
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 3e-05,
+      "loss": 0.2292,
+      "step": 1306
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 3e-05,
+      "loss": 0.28,
+      "step": 1307
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 3e-05,
+      "loss": 0.283,
+      "step": 1308
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 3e-05,
+      "loss": 0.4312,
+      "step": 1309
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 3e-05,
+      "loss": 0.1599,
+      "step": 1310
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 3e-05,
+      "loss": 0.2551,
+      "step": 1311
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 3e-05,
+      "loss": 0.3828,
+      "step": 1312
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 3e-05,
+      "loss": 0.5171,
+      "step": 1313
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 3e-05,
+      "loss": 0.2732,
+      "step": 1314
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 3e-05,
+      "loss": 0.2908,
+      "step": 1315
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 3e-05,
+      "loss": 0.4192,
+      "step": 1316
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 3e-05,
+      "loss": 0.2478,
+      "step": 1317
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1998,
+      "step": 1318
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 3e-05,
+      "loss": 0.1981,
+      "step": 1319
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 3e-05,
+      "loss": 0.4072,
+      "step": 1320
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 3e-05,
+      "loss": 0.2598,
+      "step": 1321
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 3e-05,
+      "loss": 0.2627,
+      "step": 1322
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 3e-05,
+      "loss": 0.3701,
+      "step": 1323
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 3e-05,
+      "loss": 0.3625,
+      "step": 1324
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 3e-05,
+      "loss": 0.2859,
+      "step": 1325
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 3e-05,
+      "loss": 0.3899,
+      "step": 1326
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 3e-05,
+      "loss": 0.2791,
+      "step": 1327
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 3e-05,
+      "loss": 0.2939,
+      "step": 1328
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 3e-05,
+      "loss": 0.2859,
+      "step": 1329
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 3e-05,
+      "loss": 0.2649,
+      "step": 1330
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 3e-05,
+      "loss": 0.3049,
+      "step": 1331
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 3e-05,
+      "loss": 0.3152,
+      "step": 1332
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 3e-05,
+      "loss": 0.3606,
+      "step": 1333
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 3e-05,
+      "loss": 0.2664,
+      "step": 1334
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 3e-05,
+      "loss": 0.2637,
+      "step": 1335
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 3e-05,
+      "loss": 0.3169,
+      "step": 1336
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 3e-05,
+      "loss": 0.2949,
+      "step": 1337
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 3e-05,
+      "loss": 0.7178,
+      "step": 1338
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 3e-05,
+      "loss": 0.314,
+      "step": 1339
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 3e-05,
+      "loss": 0.272,
+      "step": 1340
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 3e-05,
+      "loss": 0.2412,
+      "step": 1341
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 3e-05,
+      "loss": 0.3042,
+      "step": 1342
+    },
+    {
+      "epoch": 11.0,
+      "eval_loss": 0.332275390625,
+      "eval_mae": 0.43408203125,
+      "eval_mse": 0.332275390625,
+      "eval_runtime": 51.1414,
+      "eval_samples_per_second": 3234.032,
+      "eval_steps_per_second": 0.274,
+      "step": 1342
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 3e-05,
+      "loss": 0.2505,
+      "step": 1343
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 3e-05,
+      "loss": 0.3074,
+      "step": 1344
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 3e-05,
+      "loss": 0.2607,
+      "step": 1345
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 3e-05,
+      "loss": 0.3584,
+      "step": 1346
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 3e-05,
+      "loss": 0.245,
+      "step": 1347
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 3e-05,
+      "loss": 0.2157,
+      "step": 1348
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 3e-05,
+      "loss": 0.3154,
+      "step": 1349
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 3e-05,
+      "loss": 0.3125,
+      "step": 1350
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 3e-05,
+      "loss": 0.3413,
+      "step": 1351
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 3e-05,
+      "loss": 0.2169,
+      "step": 1352
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 3e-05,
+      "loss": 0.2448,
+      "step": 1353
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 3e-05,
+      "loss": 0.1519,
+      "step": 1354
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1917,
+      "step": 1355
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 3e-05,
+      "loss": 0.3323,
+      "step": 1356
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 3e-05,
+      "loss": 0.2634,
+      "step": 1357
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 3e-05,
+      "loss": 0.2439,
+      "step": 1358
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 3e-05,
+      "loss": 0.3252,
+      "step": 1359
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 3e-05,
+      "loss": 0.4001,
+      "step": 1360
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 3e-05,
+      "loss": 0.194,
+      "step": 1361
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 3e-05,
+      "loss": 0.3777,
+      "step": 1362
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 3e-05,
+      "loss": 0.2367,
+      "step": 1363
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 3e-05,
+      "loss": 0.3296,
+      "step": 1364
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 3e-05,
+      "loss": 0.3152,
+      "step": 1365
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 3e-05,
+      "loss": 0.2644,
+      "step": 1366
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 3e-05,
+      "loss": 0.3057,
+      "step": 1367
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 3e-05,
+      "loss": 0.4412,
+      "step": 1368
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 3e-05,
+      "loss": 0.2683,
+      "step": 1369
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 3e-05,
+      "loss": 0.2079,
+      "step": 1370
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 3e-05,
+      "loss": 0.3384,
+      "step": 1371
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 3e-05,
+      "loss": 0.364,
+      "step": 1372
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 3e-05,
+      "loss": 0.2915,
+      "step": 1373
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 3e-05,
+      "loss": 0.269,
+      "step": 1374
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 3e-05,
+      "loss": 0.2385,
+      "step": 1375
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 3e-05,
+      "loss": 0.2561,
+      "step": 1376
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 3e-05,
+      "loss": 0.2318,
+      "step": 1377
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 3e-05,
+      "loss": 0.3562,
+      "step": 1378
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 3e-05,
+      "loss": 0.3037,
+      "step": 1379
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 3e-05,
+      "loss": 0.3157,
+      "step": 1380
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 3e-05,
+      "loss": 0.2491,
+      "step": 1381
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 3e-05,
+      "loss": 0.4143,
+      "step": 1382
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 3e-05,
+      "loss": 0.2454,
+      "step": 1383
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 3e-05,
+      "loss": 0.3108,
+      "step": 1384
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 3e-05,
+      "loss": 0.2566,
+      "step": 1385
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 3e-05,
+      "loss": 0.2563,
+      "step": 1386
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 3e-05,
+      "loss": 0.1519,
+      "step": 1387
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 3e-05,
+      "loss": 0.2438,
+      "step": 1388
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1824,
+      "step": 1389
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 3e-05,
+      "loss": 0.2227,
+      "step": 1390
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 3e-05,
+      "loss": 0.2344,
+      "step": 1391
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 3e-05,
+      "loss": 0.3411,
+      "step": 1392
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 3e-05,
+      "loss": 0.3018,
+      "step": 1393
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 3e-05,
+      "loss": 0.2673,
+      "step": 1394
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 3e-05,
+      "loss": 0.282,
+      "step": 1395
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 3e-05,
+      "loss": 0.3389,
+      "step": 1396
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 3e-05,
+      "loss": 0.2197,
+      "step": 1397
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 3e-05,
+      "loss": 0.3645,
+      "step": 1398
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 3e-05,
+      "loss": 0.1956,
+      "step": 1399
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 3e-05,
+      "loss": 0.2976,
+      "step": 1400
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 3e-05,
+      "loss": 0.2842,
+      "step": 1401
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 3e-05,
+      "loss": 0.3726,
+      "step": 1402
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 3e-05,
+      "loss": 0.1975,
+      "step": 1403
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 3e-05,
+      "loss": 0.2244,
+      "step": 1404
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 3e-05,
+      "loss": 0.2362,
+      "step": 1405
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 3e-05,
+      "loss": 0.3064,
+      "step": 1406
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 3e-05,
+      "loss": 0.3308,
+      "step": 1407
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 3e-05,
+      "loss": 0.2708,
+      "step": 1408
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 3e-05,
+      "loss": 0.2627,
+      "step": 1409
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 3e-05,
+      "loss": 0.2255,
+      "step": 1410
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 3e-05,
+      "loss": 0.3003,
+      "step": 1411
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 3e-05,
+      "loss": 0.2808,
+      "step": 1412
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 3e-05,
+      "loss": 0.3643,
+      "step": 1413
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 3e-05,
+      "loss": 0.212,
+      "step": 1414
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 3e-05,
+      "loss": 0.2263,
+      "step": 1415
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 3e-05,
+      "loss": 0.2067,
+      "step": 1416
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 3e-05,
+      "loss": 0.2612,
+      "step": 1417
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 3e-05,
+      "loss": 0.2935,
+      "step": 1418
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 3e-05,
+      "loss": 0.3765,
+      "step": 1419
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 3e-05,
+      "loss": 0.4087,
+      "step": 1420
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 3e-05,
+      "loss": 0.3213,
+      "step": 1421
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 3e-05,
+      "loss": 0.2219,
+      "step": 1422
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 3e-05,
+      "loss": 0.2469,
+      "step": 1423
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 3e-05,
+      "loss": 0.245,
+      "step": 1424
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 3e-05,
+      "loss": 0.6743,
+      "step": 1425
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 3e-05,
+      "loss": 0.1943,
+      "step": 1426
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 3e-05,
+      "loss": 0.252,
+      "step": 1427
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 3e-05,
+      "loss": 0.2905,
+      "step": 1428
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 3e-05,
+      "loss": 0.2424,
+      "step": 1429
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 3e-05,
+      "loss": 0.3557,
+      "step": 1430
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 3e-05,
+      "loss": 0.2295,
+      "step": 1431
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 3e-05,
+      "loss": 0.2815,
+      "step": 1432
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 3e-05,
+      "loss": 0.2786,
+      "step": 1433
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 3e-05,
+      "loss": 0.3853,
+      "step": 1434
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 3e-05,
+      "loss": 0.415,
+      "step": 1435
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 3e-05,
+      "loss": 0.3467,
+      "step": 1436
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 3e-05,
+      "loss": 0.1924,
+      "step": 1437
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 3e-05,
+      "loss": 0.2771,
+      "step": 1438
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 3e-05,
+      "loss": 0.259,
+      "step": 1439
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 3e-05,
+      "loss": 0.3542,
+      "step": 1440
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 3e-05,
+      "loss": 0.1888,
+      "step": 1441
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 3e-05,
+      "loss": 0.2494,
+      "step": 1442
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 3e-05,
+      "loss": 0.2271,
+      "step": 1443
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 3e-05,
+      "loss": 0.2764,
+      "step": 1444
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 3e-05,
+      "loss": 0.2913,
+      "step": 1445
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 3e-05,
+      "loss": 0.2314,
+      "step": 1446
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 3e-05,
+      "loss": 0.3125,
+      "step": 1447
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 3e-05,
+      "loss": 0.3347,
+      "step": 1448
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 3e-05,
+      "loss": 0.377,
+      "step": 1449
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 3e-05,
+      "loss": 0.2278,
+      "step": 1450
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 3e-05,
+      "loss": 0.5225,
+      "step": 1451
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 3e-05,
+      "loss": 0.2404,
+      "step": 1452
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 3e-05,
+      "loss": 0.1646,
+      "step": 1453
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 3e-05,
+      "loss": 0.2206,
+      "step": 1454
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 3e-05,
+      "loss": 0.2925,
+      "step": 1455
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 3e-05,
+      "loss": 0.3069,
+      "step": 1456
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 3e-05,
+      "loss": 0.2556,
+      "step": 1457
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 3e-05,
+      "loss": 0.3479,
+      "step": 1458
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 3e-05,
+      "loss": 0.2869,
+      "step": 1459
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 3e-05,
+      "loss": 0.3467,
+      "step": 1460
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 3e-05,
+      "loss": 0.3772,
+      "step": 1461
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 3e-05,
+      "loss": 0.3093,
+      "step": 1462
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 3e-05,
+      "loss": 0.2673,
+      "step": 1463
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 3e-05,
+      "loss": 0.2211,
+      "step": 1464
+    },
+    {
+      "epoch": 12.0,
+      "eval_loss": 0.325439453125,
+      "eval_mae": 0.42919921875,
+      "eval_mse": 0.325439453125,
+      "eval_runtime": 51.111,
+      "eval_samples_per_second": 3235.956,
+      "eval_steps_per_second": 0.274,
+      "step": 1464
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 3e-05,
+      "loss": 0.3181,
+      "step": 1465
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 3e-05,
+      "loss": 0.3174,
+      "step": 1466
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 3e-05,
+      "loss": 0.2448,
+      "step": 1467
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 3e-05,
+      "loss": 0.2585,
+      "step": 1468
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 3e-05,
+      "loss": 0.2583,
+      "step": 1469
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 3e-05,
+      "loss": 0.3044,
+      "step": 1470
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 3e-05,
+      "loss": 0.4885,
+      "step": 1471
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 3e-05,
+      "loss": 0.3435,
+      "step": 1472
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 3e-05,
+      "loss": 0.2454,
+      "step": 1473
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 3e-05,
+      "loss": 0.3357,
+      "step": 1474
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 3e-05,
+      "loss": 0.239,
+      "step": 1475
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 3e-05,
+      "loss": 0.2603,
+      "step": 1476
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 3e-05,
+      "loss": 0.3188,
+      "step": 1477
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 3e-05,
+      "loss": 0.2473,
+      "step": 1478
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 3e-05,
+      "loss": 0.2294,
+      "step": 1479
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 3e-05,
+      "loss": 0.4944,
+      "step": 1480
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 3e-05,
+      "loss": 0.2925,
+      "step": 1481
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 3e-05,
+      "loss": 0.3025,
+      "step": 1482
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 3e-05,
+      "loss": 0.25,
+      "step": 1483
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 3e-05,
+      "loss": 0.2981,
+      "step": 1484
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 3e-05,
+      "loss": 0.2793,
+      "step": 1485
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 3e-05,
+      "loss": 0.4236,
+      "step": 1486
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 3e-05,
+      "loss": 0.3804,
+      "step": 1487
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 3e-05,
+      "loss": 0.2534,
+      "step": 1488
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1714,
+      "step": 1489
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 3e-05,
+      "loss": 0.2206,
+      "step": 1490
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 3e-05,
+      "loss": 0.259,
+      "step": 1491
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 3e-05,
+      "loss": 0.2355,
+      "step": 1492
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 3e-05,
+      "loss": 0.2935,
+      "step": 1493
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 3e-05,
+      "loss": 0.2981,
+      "step": 1494
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 3e-05,
+      "loss": 0.2542,
+      "step": 1495
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 3e-05,
+      "loss": 0.2512,
+      "step": 1496
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 3e-05,
+      "loss": 0.2761,
+      "step": 1497
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 3e-05,
+      "loss": 0.3977,
+      "step": 1498
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 3e-05,
+      "loss": 0.2642,
+      "step": 1499
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 3e-05,
+      "loss": 0.2798,
+      "step": 1500
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 3e-05,
+      "loss": 0.3777,
+      "step": 1501
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 3e-05,
+      "loss": 0.3584,
+      "step": 1502
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 3e-05,
+      "loss": 0.3774,
+      "step": 1503
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 3e-05,
+      "loss": 0.1973,
+      "step": 1504
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 3e-05,
+      "loss": 0.2339,
+      "step": 1505
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 3e-05,
+      "loss": 0.2036,
+      "step": 1506
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 3e-05,
+      "loss": 0.2563,
+      "step": 1507
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 3e-05,
+      "loss": 0.3088,
+      "step": 1508
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 3e-05,
+      "loss": 0.4031,
+      "step": 1509
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 3e-05,
+      "loss": 0.2173,
+      "step": 1510
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 3e-05,
+      "loss": 0.2151,
+      "step": 1511
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 3e-05,
+      "loss": 0.3481,
+      "step": 1512
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 3e-05,
+      "loss": 0.3433,
+      "step": 1513
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 3e-05,
+      "loss": 0.3618,
+      "step": 1514
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 3e-05,
+      "loss": 0.2712,
+      "step": 1515
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 3e-05,
+      "loss": 0.2034,
+      "step": 1516
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 3e-05,
+      "loss": 0.2874,
+      "step": 1517
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 3e-05,
+      "loss": 0.157,
+      "step": 1518
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 3e-05,
+      "loss": 0.2554,
+      "step": 1519
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 3e-05,
+      "loss": 0.3042,
+      "step": 1520
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 3e-05,
+      "loss": 0.4397,
+      "step": 1521
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 3e-05,
+      "loss": 0.248,
+      "step": 1522
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 3e-05,
+      "loss": 0.3516,
+      "step": 1523
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 3e-05,
+      "loss": 0.3167,
+      "step": 1524
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 3e-05,
+      "loss": 0.3037,
+      "step": 1525
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 3e-05,
+      "loss": 0.2284,
+      "step": 1526
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 3e-05,
+      "loss": 0.3401,
+      "step": 1527
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 3e-05,
+      "loss": 0.2532,
+      "step": 1528
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 3e-05,
+      "loss": 0.3296,
+      "step": 1529
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 3e-05,
+      "loss": 0.2102,
+      "step": 1530
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 3e-05,
+      "loss": 0.3403,
+      "step": 1531
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 3e-05,
+      "loss": 0.259,
+      "step": 1532
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 3e-05,
+      "loss": 0.1646,
+      "step": 1533
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 3e-05,
+      "loss": 0.498,
+      "step": 1534
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 3e-05,
+      "loss": 0.3518,
+      "step": 1535
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 3e-05,
+      "loss": 0.2324,
+      "step": 1536
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 3e-05,
+      "loss": 0.3362,
+      "step": 1537
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 3e-05,
+      "loss": 0.2915,
+      "step": 1538
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 3e-05,
+      "loss": 0.2025,
+      "step": 1539
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 3e-05,
+      "loss": 0.2332,
+      "step": 1540
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 3e-05,
+      "loss": 0.2096,
+      "step": 1541
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 3e-05,
+      "loss": 0.2715,
+      "step": 1542
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 3e-05,
+      "loss": 0.3159,
+      "step": 1543
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 3e-05,
+      "loss": 0.239,
+      "step": 1544
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 3e-05,
+      "loss": 0.2188,
+      "step": 1545
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 3e-05,
+      "loss": 0.2343,
+      "step": 1546
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 3e-05,
+      "loss": 0.1897,
+      "step": 1547
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 3e-05,
+      "loss": 0.2477,
+      "step": 1548
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 3e-05,
+      "loss": 0.2103,
+      "step": 1549
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 3e-05,
+      "loss": 0.2169,
+      "step": 1550
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 3e-05,
+      "loss": 0.2698,
+      "step": 1551
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 3e-05,
+      "loss": 0.282,
+      "step": 1552
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 3e-05,
+      "loss": 0.251,
+      "step": 1553
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 3e-05,
+      "loss": 0.3752,
+      "step": 1554
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 3e-05,
+      "loss": 0.3091,
+      "step": 1555
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 3e-05,
+      "loss": 0.3484,
+      "step": 1556
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 3e-05,
+      "loss": 0.2607,
+      "step": 1557
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 3e-05,
+      "loss": 0.2766,
+      "step": 1558
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 3e-05,
+      "loss": 0.3445,
+      "step": 1559
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 3e-05,
+      "loss": 0.1987,
+      "step": 1560
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 3e-05,
+      "loss": 0.2734,
+      "step": 1561
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 3e-05,
+      "loss": 0.2861,
+      "step": 1562
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 3e-05,
+      "loss": 0.406,
+      "step": 1563
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 3e-05,
+      "loss": 0.2937,
+      "step": 1564
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 3e-05,
+      "loss": 0.2939,
+      "step": 1565
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 3e-05,
+      "loss": 0.2542,
+      "step": 1566
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 3e-05,
+      "loss": 0.271,
+      "step": 1567
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 3e-05,
+      "loss": 0.235,
+      "step": 1568
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 3e-05,
+      "loss": 0.292,
+      "step": 1569
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 3e-05,
+      "loss": 0.2064,
+      "step": 1570
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 3e-05,
+      "loss": 0.2448,
+      "step": 1571
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 3e-05,
+      "loss": 0.426,
+      "step": 1572
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 3e-05,
+      "loss": 0.2206,
+      "step": 1573
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 3e-05,
+      "loss": 0.2289,
+      "step": 1574
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 3e-05,
+      "loss": 0.3149,
+      "step": 1575
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 3e-05,
+      "loss": 0.2947,
+      "step": 1576
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 3e-05,
+      "loss": 0.2678,
+      "step": 1577
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 3e-05,
+      "loss": 0.2249,
+      "step": 1578
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 3e-05,
+      "loss": 0.2666,
+      "step": 1579
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 3e-05,
+      "loss": 0.3586,
+      "step": 1580
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 3e-05,
+      "loss": 0.2417,
+      "step": 1581
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 3e-05,
+      "loss": 0.3723,
+      "step": 1582
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 3e-05,
+      "loss": 0.3333,
+      "step": 1583
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 3e-05,
+      "loss": 0.3953,
+      "step": 1584
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 3e-05,
+      "loss": 0.212,
+      "step": 1585
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 3e-05,
+      "loss": 0.6309,
+      "step": 1586
+    },
+    {
+      "epoch": 13.0,
+      "eval_loss": 0.32666015625,
+      "eval_mae": 0.431640625,
+      "eval_mse": 0.32666015625,
+      "eval_runtime": 51.1163,
+      "eval_samples_per_second": 3235.624,
+      "eval_steps_per_second": 0.274,
+      "step": 1586
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 3e-05,
+      "loss": 0.3108,
+      "step": 1587
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 3e-05,
+      "loss": 0.4536,
+      "step": 1588
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 3e-05,
+      "loss": 0.3721,
+      "step": 1589
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 3e-05,
+      "loss": 0.2166,
+      "step": 1590
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 3e-05,
+      "loss": 0.2373,
+      "step": 1591
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 3e-05,
+      "loss": 0.2207,
+      "step": 1592
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 3e-05,
+      "loss": 0.2642,
+      "step": 1593
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 3e-05,
+      "loss": 0.2805,
+      "step": 1594
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 3e-05,
+      "loss": 0.2262,
+      "step": 1595
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 3e-05,
+      "loss": 0.2744,
+      "step": 1596
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 3e-05,
+      "loss": 0.2277,
+      "step": 1597
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 3e-05,
+      "loss": 0.2732,
+      "step": 1598
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 3e-05,
+      "loss": 0.4114,
+      "step": 1599
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1936,
+      "step": 1600
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 3e-05,
+      "loss": 0.3167,
+      "step": 1601
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 3e-05,
+      "loss": 0.1666,
+      "step": 1602
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 3e-05,
+      "loss": 0.3857,
+      "step": 1603
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 3e-05,
+      "loss": 0.2925,
+      "step": 1604
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 3e-05,
+      "loss": 0.2137,
+      "step": 1605
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 3e-05,
+      "loss": 0.292,
+      "step": 1606
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 3e-05,
+      "loss": 0.2805,
+      "step": 1607
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 3e-05,
+      "loss": 0.3354,
+      "step": 1608
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 3e-05,
+      "loss": 0.2159,
+      "step": 1609
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 3e-05,
+      "loss": 0.3542,
+      "step": 1610
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 3e-05,
+      "loss": 0.3213,
+      "step": 1611
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 3e-05,
+      "loss": 0.2585,
+      "step": 1612
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 3e-05,
+      "loss": 0.3328,
+      "step": 1613
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 3e-05,
+      "loss": 0.2688,
+      "step": 1614
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 3e-05,
+      "loss": 0.4199,
+      "step": 1615
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 3e-05,
+      "loss": 0.3682,
+      "step": 1616
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 3e-05,
+      "loss": 0.2815,
+      "step": 1617
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 3e-05,
+      "loss": 0.1946,
+      "step": 1618
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 3e-05,
+      "loss": 0.3059,
+      "step": 1619
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 3e-05,
+      "loss": 0.2668,
+      "step": 1620
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 3e-05,
+      "loss": 0.2157,
+      "step": 1621
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 3e-05,
+      "loss": 0.2568,
+      "step": 1622
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 3e-05,
+      "loss": 0.293,
+      "step": 1623
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 3e-05,
+      "loss": 0.1877,
+      "step": 1624
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 3e-05,
+      "loss": 0.2456,
+      "step": 1625
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 3e-05,
+      "loss": 0.3467,
+      "step": 1626
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1639,
+      "step": 1627
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 3e-05,
+      "loss": 0.2415,
+      "step": 1628
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 3e-05,
+      "loss": 0.282,
+      "step": 1629
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 3e-05,
+      "loss": 0.3254,
+      "step": 1630
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 3e-05,
+      "loss": 0.2654,
+      "step": 1631
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 3e-05,
+      "loss": 0.2998,
+      "step": 1632
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1702,
+      "step": 1633
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 3e-05,
+      "loss": 0.3577,
+      "step": 1634
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 3e-05,
+      "loss": 0.2866,
+      "step": 1635
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 3e-05,
+      "loss": 0.2563,
+      "step": 1636
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 3e-05,
+      "loss": 0.2844,
+      "step": 1637
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 3e-05,
+      "loss": 0.2864,
+      "step": 1638
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 3e-05,
+      "loss": 0.3494,
+      "step": 1639
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 3e-05,
+      "loss": 0.2903,
+      "step": 1640
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 3e-05,
+      "loss": 0.2537,
+      "step": 1641
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 3e-05,
+      "loss": 0.2791,
+      "step": 1642
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 3e-05,
+      "loss": 0.2366,
+      "step": 1643
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 3e-05,
+      "loss": 0.2874,
+      "step": 1644
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 3e-05,
+      "loss": 0.2153,
+      "step": 1645
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 3e-05,
+      "loss": 0.2321,
+      "step": 1646
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 3e-05,
+      "loss": 0.3132,
+      "step": 1647
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 3e-05,
+      "loss": 0.3445,
+      "step": 1648
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 3e-05,
+      "loss": 0.2817,
+      "step": 1649
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 3e-05,
+      "loss": 0.215,
+      "step": 1650
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 3e-05,
+      "loss": 0.1993,
+      "step": 1651
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 3e-05,
+      "loss": 0.2472,
+      "step": 1652
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 3e-05,
+      "loss": 0.2427,
+      "step": 1653
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 3e-05,
+      "loss": 0.3694,
+      "step": 1654
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 3e-05,
+      "loss": 0.1606,
+      "step": 1655
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 3e-05,
+      "loss": 0.2301,
+      "step": 1656
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 3e-05,
+      "loss": 0.2769,
+      "step": 1657
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 3e-05,
+      "loss": 0.2191,
+      "step": 1658
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 3e-05,
+      "loss": 0.2778,
+      "step": 1659
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 3e-05,
+      "loss": 0.2988,
+      "step": 1660
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 3e-05,
+      "loss": 0.3447,
+      "step": 1661
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 3e-05,
+      "loss": 0.2688,
+      "step": 1662
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 3e-05,
+      "loss": 0.3533,
+      "step": 1663
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 3e-05,
+      "loss": 0.4675,
+      "step": 1664
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 3e-05,
+      "loss": 0.3149,
+      "step": 1665
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 3e-05,
+      "loss": 0.3442,
+      "step": 1666
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 3e-05,
+      "loss": 0.3176,
+      "step": 1667
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 3e-05,
+      "loss": 0.2654,
+      "step": 1668
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 3e-05,
+      "loss": 0.3103,
+      "step": 1669
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 3e-05,
+      "loss": 0.2551,
+      "step": 1670
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 3e-05,
+      "loss": 0.3074,
+      "step": 1671
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 3e-05,
+      "loss": 0.4207,
+      "step": 1672
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 3e-05,
+      "loss": 0.2216,
+      "step": 1673
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 3e-05,
+      "loss": 0.2424,
+      "step": 1674
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 3e-05,
+      "loss": 0.2798,
+      "step": 1675
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 3e-05,
+      "loss": 0.2512,
+      "step": 1676
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 3e-05,
+      "loss": 0.2888,
+      "step": 1677
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 3e-05,
+      "loss": 0.2988,
+      "step": 1678
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 3e-05,
+      "loss": 0.3462,
+      "step": 1679
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 3e-05,
+      "loss": 0.3506,
+      "step": 1680
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 3e-05,
+      "loss": 0.1986,
+      "step": 1681
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 3e-05,
+      "loss": 0.2334,
+      "step": 1682
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 3e-05,
+      "loss": 0.215,
+      "step": 1683
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 3e-05,
+      "loss": 0.2939,
+      "step": 1684
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 3e-05,
+      "loss": 0.2272,
+      "step": 1685
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 3e-05,
+      "loss": 0.1753,
+      "step": 1686
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 3e-05,
+      "loss": 0.2954,
+      "step": 1687
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 3e-05,
+      "loss": 0.3071,
+      "step": 1688
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1864,
+      "step": 1689
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 3e-05,
+      "loss": 0.2827,
+      "step": 1690
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 3e-05,
+      "loss": 0.27,
+      "step": 1691
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 3e-05,
+      "loss": 0.333,
+      "step": 1692
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 3e-05,
+      "loss": 0.3364,
+      "step": 1693
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 3e-05,
+      "loss": 0.4185,
+      "step": 1694
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1635,
+      "step": 1695
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 3e-05,
+      "loss": 0.1648,
+      "step": 1696
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 3e-05,
+      "loss": 0.2874,
+      "step": 1697
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 3e-05,
+      "loss": 0.1696,
+      "step": 1698
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 3e-05,
+      "loss": 0.3616,
+      "step": 1699
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 3e-05,
+      "loss": 0.2625,
+      "step": 1700
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 3e-05,
+      "loss": 0.2605,
+      "step": 1701
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 3e-05,
+      "loss": 0.3289,
+      "step": 1702
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 3e-05,
+      "loss": 0.2576,
+      "step": 1703
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 3e-05,
+      "loss": 0.2211,
+      "step": 1704
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 3e-05,
+      "loss": 0.3259,
+      "step": 1705
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 3e-05,
+      "loss": 0.2749,
+      "step": 1706
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 3e-05,
+      "loss": 0.3074,
+      "step": 1707
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3e-05,
+      "loss": 0.1849,
+      "step": 1708
+    },
+    {
+      "epoch": 14.0,
+      "eval_loss": 0.310546875,
+      "eval_mae": 0.41650390625,
+      "eval_mse": 0.310546875,
+      "eval_runtime": 51.0451,
+      "eval_samples_per_second": 3240.132,
+      "eval_steps_per_second": 0.274,
+      "step": 1708
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3e-05,
+      "loss": 0.2678,
+      "step": 1709
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3e-05,
+      "loss": 0.2496,
+      "step": 1710
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3e-05,
+      "loss": 0.2292,
+      "step": 1711
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3e-05,
+      "loss": 0.2341,
+      "step": 1712
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3e-05,
+      "loss": 0.2686,
+      "step": 1713
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3e-05,
+      "loss": 0.1962,
+      "step": 1714
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3e-05,
+      "loss": 0.243,
+      "step": 1715
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3e-05,
+      "loss": 0.2539,
+      "step": 1716
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1929,
+      "step": 1717
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3e-05,
+      "loss": 0.2942,
+      "step": 1718
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3e-05,
+      "loss": 0.2668,
+      "step": 1719
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3e-05,
+      "loss": 0.1742,
+      "step": 1720
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3e-05,
+      "loss": 0.2457,
+      "step": 1721
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3e-05,
+      "loss": 0.3262,
+      "step": 1722
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3e-05,
+      "loss": 0.3181,
+      "step": 1723
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3e-05,
+      "loss": 0.2798,
+      "step": 1724
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3e-05,
+      "loss": 0.1816,
+      "step": 1725
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3e-05,
+      "loss": 0.3577,
+      "step": 1726
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3e-05,
+      "loss": 0.2075,
+      "step": 1727
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3e-05,
+      "loss": 0.3125,
+      "step": 1728
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3e-05,
+      "loss": 0.1733,
+      "step": 1729
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3e-05,
+      "loss": 0.2323,
+      "step": 1730
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3e-05,
+      "loss": 0.1913,
+      "step": 1731
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3e-05,
+      "loss": 0.375,
+      "step": 1732
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3e-05,
+      "loss": 0.2751,
+      "step": 1733
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3e-05,
+      "loss": 0.271,
+      "step": 1734
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3e-05,
+      "loss": 0.24,
+      "step": 1735
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3e-05,
+      "loss": 0.2479,
+      "step": 1736
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3e-05,
+      "loss": 0.2245,
+      "step": 1737
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3e-05,
+      "loss": 0.2253,
+      "step": 1738
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1591,
+      "step": 1739
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3e-05,
+      "loss": 0.2549,
+      "step": 1740
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3e-05,
+      "loss": 0.2468,
+      "step": 1741
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3e-05,
+      "loss": 0.2522,
+      "step": 1742
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3e-05,
+      "loss": 0.3035,
+      "step": 1743
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1632,
+      "step": 1744
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3e-05,
+      "loss": 0.2859,
+      "step": 1745
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3e-05,
+      "loss": 0.2307,
+      "step": 1746
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3e-05,
+      "loss": 0.2283,
+      "step": 1747
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3e-05,
+      "loss": 0.214,
+      "step": 1748
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3e-05,
+      "loss": 0.2588,
+      "step": 1749
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1785,
+      "step": 1750
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 3e-05,
+      "loss": 0.2241,
+      "step": 1751
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 3e-05,
+      "loss": 0.3049,
+      "step": 1752
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 3e-05,
+      "loss": 0.3711,
+      "step": 1753
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3e-05,
+      "loss": 0.2725,
+      "step": 1754
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3e-05,
+      "loss": 0.2406,
+      "step": 1755
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1919,
+      "step": 1756
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 3e-05,
+      "loss": 0.3594,
+      "step": 1757
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3e-05,
+      "loss": 0.5669,
+      "step": 1758
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3e-05,
+      "loss": 0.1833,
+      "step": 1759
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3e-05,
+      "loss": 0.2568,
+      "step": 1760
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3e-05,
+      "loss": 0.3008,
+      "step": 1761
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3e-05,
+      "loss": 0.2371,
+      "step": 1762
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3e-05,
+      "loss": 0.3528,
+      "step": 1763
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3e-05,
+      "loss": 0.3032,
+      "step": 1764
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3e-05,
+      "loss": 0.2406,
+      "step": 1765
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3e-05,
+      "loss": 0.2045,
+      "step": 1766
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3e-05,
+      "loss": 0.208,
+      "step": 1767
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3e-05,
+      "loss": 0.1869,
+      "step": 1768
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3e-05,
+      "loss": 0.2456,
+      "step": 1769
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3e-05,
+      "loss": 0.2959,
+      "step": 1770
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1931,
+      "step": 1771
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3e-05,
+      "loss": 0.2896,
+      "step": 1772
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3e-05,
+      "loss": 0.2874,
+      "step": 1773
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3e-05,
+      "loss": 0.262,
+      "step": 1774
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3e-05,
+      "loss": 0.3228,
+      "step": 1775
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 3e-05,
+      "loss": 0.2991,
+      "step": 1776
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3e-05,
+      "loss": 0.2079,
+      "step": 1777
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3e-05,
+      "loss": 0.2203,
+      "step": 1778
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3e-05,
+      "loss": 0.1838,
+      "step": 1779
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3e-05,
+      "loss": 0.188,
+      "step": 1780
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3e-05,
+      "loss": 0.2474,
+      "step": 1781
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3e-05,
+      "loss": 0.2324,
+      "step": 1782
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3e-05,
+      "loss": 0.2267,
+      "step": 1783
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3e-05,
+      "loss": 0.3083,
+      "step": 1784
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3e-05,
+      "loss": 0.2078,
+      "step": 1785
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3e-05,
+      "loss": 0.8315,
+      "step": 1786
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3e-05,
+      "loss": 0.2637,
+      "step": 1787
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3e-05,
+      "loss": 0.2698,
+      "step": 1788
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3e-05,
+      "loss": 0.2869,
+      "step": 1789
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3e-05,
+      "loss": 0.3115,
+      "step": 1790
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3e-05,
+      "loss": 0.5483,
+      "step": 1791
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 3e-05,
+      "loss": 0.1594,
+      "step": 1792
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3e-05,
+      "loss": 0.2688,
+      "step": 1793
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3e-05,
+      "loss": 0.345,
+      "step": 1794
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3e-05,
+      "loss": 0.3748,
+      "step": 1795
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3e-05,
+      "loss": 0.282,
+      "step": 1796
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3e-05,
+      "loss": 0.2515,
+      "step": 1797
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3e-05,
+      "loss": 0.1864,
+      "step": 1798
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3e-05,
+      "loss": 0.3054,
+      "step": 1799
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3e-05,
+      "loss": 0.2935,
+      "step": 1800
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3e-05,
+      "loss": 0.2494,
+      "step": 1801
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3e-05,
+      "loss": 0.213,
+      "step": 1802
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3e-05,
+      "loss": 0.2419,
+      "step": 1803
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 3e-05,
+      "loss": 0.3247,
+      "step": 1804
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3e-05,
+      "loss": 0.2296,
+      "step": 1805
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3e-05,
+      "loss": 0.281,
+      "step": 1806
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3e-05,
+      "loss": 0.2754,
+      "step": 1807
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 3e-05,
+      "loss": 0.3018,
+      "step": 1808
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 3e-05,
+      "loss": 0.1984,
+      "step": 1809
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3e-05,
+      "loss": 0.2502,
+      "step": 1810
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3e-05,
+      "loss": 0.2534,
+      "step": 1811
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 3e-05,
+      "loss": 0.2744,
+      "step": 1812
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 3e-05,
+      "loss": 0.1694,
+      "step": 1813
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3e-05,
+      "loss": 0.2073,
+      "step": 1814
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3e-05,
+      "loss": 0.3076,
+      "step": 1815
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3e-05,
+      "loss": 0.3096,
+      "step": 1816
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1943,
+      "step": 1817
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3e-05,
+      "loss": 0.2107,
+      "step": 1818
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3e-05,
+      "loss": 0.2664,
+      "step": 1819
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 3e-05,
+      "loss": 0.2842,
+      "step": 1820
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3e-05,
+      "loss": 0.2,
+      "step": 1821
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3e-05,
+      "loss": 0.231,
+      "step": 1822
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 3e-05,
+      "loss": 0.2406,
+      "step": 1823
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 3e-05,
+      "loss": 0.2834,
+      "step": 1824
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3e-05,
+      "loss": 0.225,
+      "step": 1825
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 3e-05,
+      "loss": 0.3677,
+      "step": 1826
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3e-05,
+      "loss": 0.2423,
+      "step": 1827
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3e-05,
+      "loss": 0.2568,
+      "step": 1828
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 3e-05,
+      "loss": 0.2874,
+      "step": 1829
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3e-05,
+      "loss": 0.157,
+      "step": 1830
+    },
+    {
+      "epoch": 15.0,
+      "eval_loss": 0.311767578125,
+      "eval_mae": 0.4169921875,
+      "eval_mse": 0.311767578125,
+      "eval_runtime": 51.1055,
+      "eval_samples_per_second": 3236.304,
+      "eval_steps_per_second": 0.274,
+      "step": 1830
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3e-05,
+      "loss": 0.3145,
+      "step": 1831
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3e-05,
+      "loss": 0.2101,
+      "step": 1832
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1893,
+      "step": 1833
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 3e-05,
+      "loss": 0.1893,
+      "step": 1834
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3e-05,
+      "loss": 0.1989,
+      "step": 1835
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 3e-05,
+      "loss": 0.1602,
+      "step": 1836
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 3e-05,
+      "loss": 0.2524,
+      "step": 1837
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3e-05,
+      "loss": 0.2798,
+      "step": 1838
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3e-05,
+      "loss": 0.218,
+      "step": 1839
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 3e-05,
+      "loss": 0.2327,
+      "step": 1840
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 3e-05,
+      "loss": 0.2563,
+      "step": 1841
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 3e-05,
+      "loss": 0.2732,
+      "step": 1842
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3e-05,
+      "loss": 0.2281,
+      "step": 1843
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3e-05,
+      "loss": 0.2612,
+      "step": 1844
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 3e-05,
+      "loss": 0.2374,
+      "step": 1845
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 3e-05,
+      "loss": 0.2632,
+      "step": 1846
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3e-05,
+      "loss": 0.2098,
+      "step": 1847
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 3e-05,
+      "loss": 0.1613,
+      "step": 1848
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1707,
+      "step": 1849
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3e-05,
+      "loss": 0.3215,
+      "step": 1850
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3e-05,
+      "loss": 0.2715,
+      "step": 1851
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 3e-05,
+      "loss": 0.2107,
+      "step": 1852
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 3e-05,
+      "loss": 0.2452,
+      "step": 1853
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3e-05,
+      "loss": 0.2369,
+      "step": 1854
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1599,
+      "step": 1855
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3e-05,
+      "loss": 0.2979,
+      "step": 1856
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 3e-05,
+      "loss": 0.175,
+      "step": 1857
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 3e-05,
+      "loss": 0.3086,
+      "step": 1858
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3e-05,
+      "loss": 0.2332,
+      "step": 1859
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3e-05,
+      "loss": 0.2915,
+      "step": 1860
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3e-05,
+      "loss": 0.2568,
+      "step": 1861
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 3e-05,
+      "loss": 0.2681,
+      "step": 1862
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3e-05,
+      "loss": 0.3491,
+      "step": 1863
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 3e-05,
+      "loss": 0.2156,
+      "step": 1864
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 3e-05,
+      "loss": 0.5015,
+      "step": 1865
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3e-05,
+      "loss": 0.2524,
+      "step": 1866
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3e-05,
+      "loss": 0.3843,
+      "step": 1867
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3e-05,
+      "loss": 0.2744,
+      "step": 1868
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 3e-05,
+      "loss": 0.2866,
+      "step": 1869
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 3e-05,
+      "loss": 0.3508,
+      "step": 1870
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3e-05,
+      "loss": 0.3225,
+      "step": 1871
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3e-05,
+      "loss": 0.2766,
+      "step": 1872
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 3e-05,
+      "loss": 0.2529,
+      "step": 1873
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 3e-05,
+      "loss": 0.2812,
+      "step": 1874
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3e-05,
+      "loss": 0.1857,
+      "step": 1875
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 3e-05,
+      "loss": 0.3315,
+      "step": 1876
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3e-05,
+      "loss": 0.2561,
+      "step": 1877
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3e-05,
+      "loss": 0.2327,
+      "step": 1878
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3e-05,
+      "loss": 0.4565,
+      "step": 1879
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 3e-05,
+      "loss": 0.27,
+      "step": 1880
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 3e-05,
+      "loss": 0.1816,
+      "step": 1881
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 3e-05,
+      "loss": 0.291,
+      "step": 1882
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 3e-05,
+      "loss": 0.3606,
+      "step": 1883
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3e-05,
+      "loss": 0.1835,
+      "step": 1884
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 3e-05,
+      "loss": 0.1991,
+      "step": 1885
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 3e-05,
+      "loss": 0.2399,
+      "step": 1886
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 3e-05,
+      "loss": 0.1445,
+      "step": 1887
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3e-05,
+      "loss": 0.271,
+      "step": 1888
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3e-05,
+      "loss": 0.3101,
+      "step": 1889
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 3e-05,
+      "loss": 0.1591,
+      "step": 1890
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3e-05,
+      "loss": 0.2844,
+      "step": 1891
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 3e-05,
+      "loss": 0.3042,
+      "step": 1892
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 3e-05,
+      "loss": 0.2881,
+      "step": 1893
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 3e-05,
+      "loss": 0.2094,
+      "step": 1894
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 3e-05,
+      "loss": 0.1687,
+      "step": 1895
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3e-05,
+      "loss": 0.1661,
+      "step": 1896
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 3e-05,
+      "loss": 0.4258,
+      "step": 1897
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 3e-05,
+      "loss": 0.203,
+      "step": 1898
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3e-05,
+      "loss": 0.3301,
+      "step": 1899
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3e-05,
+      "loss": 0.2355,
+      "step": 1900
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3e-05,
+      "loss": 0.261,
+      "step": 1901
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3e-05,
+      "loss": 0.1665,
+      "step": 1902
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3e-05,
+      "loss": 0.1305,
+      "step": 1903
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3e-05,
+      "loss": 0.2734,
+      "step": 1904
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3e-05,
+      "loss": 0.2267,
+      "step": 1905
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 3e-05,
+      "loss": 0.2273,
+      "step": 1906
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 3e-05,
+      "loss": 0.208,
+      "step": 1907
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3e-05,
+      "loss": 0.1864,
+      "step": 1908
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 3e-05,
+      "loss": 0.4121,
+      "step": 1909
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1791,
+      "step": 1910
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3e-05,
+      "loss": 0.229,
+      "step": 1911
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 3e-05,
+      "loss": 0.1843,
+      "step": 1912
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 3e-05,
+      "loss": 0.2024,
+      "step": 1913
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 3e-05,
+      "loss": 0.3589,
+      "step": 1914
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3e-05,
+      "loss": 0.3044,
+      "step": 1915
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1504,
+      "step": 1916
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 3e-05,
+      "loss": 0.2429,
+      "step": 1917
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 3e-05,
+      "loss": 0.3047,
+      "step": 1918
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 3e-05,
+      "loss": 0.1847,
+      "step": 1919
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3e-05,
+      "loss": 0.2301,
+      "step": 1920
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3e-05,
+      "loss": 0.2832,
+      "step": 1921
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3e-05,
+      "loss": 0.2437,
+      "step": 1922
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 3e-05,
+      "loss": 0.16,
+      "step": 1923
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 3e-05,
+      "loss": 0.2151,
+      "step": 1924
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 3e-05,
+      "loss": 0.25,
+      "step": 1925
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 3e-05,
+      "loss": 0.1609,
+      "step": 1926
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3e-05,
+      "loss": 0.2686,
+      "step": 1927
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3e-05,
+      "loss": 0.2109,
+      "step": 1928
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 3e-05,
+      "loss": 0.2576,
+      "step": 1929
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 3e-05,
+      "loss": 0.3127,
+      "step": 1930
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3e-05,
+      "loss": 0.3369,
+      "step": 1931
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3e-05,
+      "loss": 0.2732,
+      "step": 1932
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3e-05,
+      "loss": 0.334,
+      "step": 1933
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 3e-05,
+      "loss": 0.5562,
+      "step": 1934
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 3e-05,
+      "loss": 0.2944,
+      "step": 1935
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3e-05,
+      "loss": 0.3533,
+      "step": 1936
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3e-05,
+      "loss": 0.2239,
+      "step": 1937
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1912,
+      "step": 1938
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1742,
+      "step": 1939
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3e-05,
+      "loss": 0.2588,
+      "step": 1940
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 3e-05,
+      "loss": 0.2554,
+      "step": 1941
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 3e-05,
+      "loss": 0.3079,
+      "step": 1942
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3e-05,
+      "loss": 0.3811,
+      "step": 1943
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3e-05,
+      "loss": 0.3071,
+      "step": 1944
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 3e-05,
+      "loss": 0.2502,
+      "step": 1945
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 3e-05,
+      "loss": 0.3511,
+      "step": 1946
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 3e-05,
+      "loss": 0.3801,
+      "step": 1947
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3e-05,
+      "loss": 0.2939,
+      "step": 1948
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3e-05,
+      "loss": 0.2896,
+      "step": 1949
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3e-05,
+      "loss": 0.2217,
+      "step": 1950
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 3e-05,
+      "loss": 0.2642,
+      "step": 1951
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3e-05,
+      "loss": 0.4287,
+      "step": 1952
+    },
+    {
+      "epoch": 16.0,
+      "eval_loss": 0.300537109375,
+      "eval_mae": 0.409423828125,
+      "eval_mse": 0.300537109375,
+      "eval_runtime": 51.0958,
+      "eval_samples_per_second": 3236.923,
+      "eval_steps_per_second": 0.274,
+      "step": 1952
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 3e-05,
+      "loss": 0.1793,
+      "step": 1953
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1942,
+      "step": 1954
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1915,
+      "step": 1955
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3e-05,
+      "loss": 0.2224,
+      "step": 1956
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 3e-05,
+      "loss": 0.248,
+      "step": 1957
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 3e-05,
+      "loss": 0.2059,
+      "step": 1958
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 3e-05,
+      "loss": 0.2494,
+      "step": 1959
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3e-05,
+      "loss": 0.3716,
+      "step": 1960
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1692,
+      "step": 1961
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 3e-05,
+      "loss": 0.2347,
+      "step": 1962
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3e-05,
+      "loss": 0.3816,
+      "step": 1963
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3e-05,
+      "loss": 0.1992,
+      "step": 1964
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1908,
+      "step": 1965
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1855,
+      "step": 1966
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 3e-05,
+      "loss": 0.2957,
+      "step": 1967
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3e-05,
+      "loss": 0.2878,
+      "step": 1968
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 3e-05,
+      "loss": 0.2786,
+      "step": 1969
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 3e-05,
+      "loss": 0.2489,
+      "step": 1970
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 3e-05,
+      "loss": 0.2817,
+      "step": 1971
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 3e-05,
+      "loss": 0.2164,
+      "step": 1972
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3e-05,
+      "loss": 0.2155,
+      "step": 1973
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 3e-05,
+      "loss": 0.2695,
+      "step": 1974
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 3e-05,
+      "loss": 0.1665,
+      "step": 1975
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3e-05,
+      "loss": 0.353,
+      "step": 1976
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3e-05,
+      "loss": 0.2242,
+      "step": 1977
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 3e-05,
+      "loss": 0.301,
+      "step": 1978
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 3e-05,
+      "loss": 0.2712,
+      "step": 1979
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3e-05,
+      "loss": 0.1323,
+      "step": 1980
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 3e-05,
+      "loss": 0.2825,
+      "step": 1981
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3e-05,
+      "loss": 0.2671,
+      "step": 1982
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3e-05,
+      "loss": 0.2042,
+      "step": 1983
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 3e-05,
+      "loss": 0.3296,
+      "step": 1984
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 3e-05,
+      "loss": 0.2739,
+      "step": 1985
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 3e-05,
+      "loss": 0.2374,
+      "step": 1986
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 3e-05,
+      "loss": 0.2849,
+      "step": 1987
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3e-05,
+      "loss": 0.3491,
+      "step": 1988
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3e-05,
+      "loss": 0.3193,
+      "step": 1989
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 3e-05,
+      "loss": 0.2456,
+      "step": 1990
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 3e-05,
+      "loss": 0.1306,
+      "step": 1991
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3e-05,
+      "loss": 0.2759,
+      "step": 1992
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3e-05,
+      "loss": 0.2256,
+      "step": 1993
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1849,
+      "step": 1994
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 3e-05,
+      "loss": 0.2091,
+      "step": 1995
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3e-05,
+      "loss": 0.3118,
+      "step": 1996
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 3e-05,
+      "loss": 0.3677,
+      "step": 1997
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 3e-05,
+      "loss": 0.2676,
+      "step": 1998
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1791,
+      "step": 1999
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3e-05,
+      "loss": 0.2045,
+      "step": 2000
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3e-05,
+      "loss": 0.3167,
+      "step": 2001
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3e-05,
+      "loss": 0.1631,
+      "step": 2002
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3e-05,
+      "loss": 0.2913,
+      "step": 2003
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3e-05,
+      "loss": 0.2074,
+      "step": 2004
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3e-05,
+      "loss": 0.2245,
+      "step": 2005
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3e-05,
+      "loss": 0.311,
+      "step": 2006
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3e-05,
+      "loss": 0.2842,
+      "step": 2007
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3e-05,
+      "loss": 0.196,
+      "step": 2008
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3e-05,
+      "loss": 0.2849,
+      "step": 2009
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1774,
+      "step": 2010
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3e-05,
+      "loss": 0.2401,
+      "step": 2011
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3e-05,
+      "loss": 0.1825,
+      "step": 2012
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3e-05,
+      "loss": 0.3237,
+      "step": 2013
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3e-05,
+      "loss": 0.25,
+      "step": 2014
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3e-05,
+      "loss": 0.2312,
+      "step": 2015
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3e-05,
+      "loss": 0.2482,
+      "step": 2016
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3e-05,
+      "loss": 0.1616,
+      "step": 2017
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3e-05,
+      "loss": 0.1741,
+      "step": 2018
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3e-05,
+      "loss": 0.2053,
+      "step": 2019
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3e-05,
+      "loss": 0.2808,
+      "step": 2020
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3e-05,
+      "loss": 0.2571,
+      "step": 2021
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3e-05,
+      "loss": 0.1862,
+      "step": 2022
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3e-05,
+      "loss": 0.2688,
+      "step": 2023
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3e-05,
+      "loss": 0.1565,
+      "step": 2024
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3e-05,
+      "loss": 0.199,
+      "step": 2025
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3e-05,
+      "loss": 0.2472,
+      "step": 2026
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3e-05,
+      "loss": 0.3091,
+      "step": 2027
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3e-05,
+      "loss": 0.1796,
+      "step": 2028
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3e-05,
+      "loss": 0.2712,
+      "step": 2029
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3e-05,
+      "loss": 0.2421,
+      "step": 2030
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3e-05,
+      "loss": 0.2423,
+      "step": 2031
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3e-05,
+      "loss": 0.2659,
+      "step": 2032
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3e-05,
+      "loss": 0.2089,
+      "step": 2033
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3e-05,
+      "loss": 0.1759,
+      "step": 2034
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3e-05,
+      "loss": 0.2457,
+      "step": 2035
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3e-05,
+      "loss": 0.2428,
+      "step": 2036
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3e-05,
+      "loss": 0.2283,
+      "step": 2037
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1501,
+      "step": 2038
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3e-05,
+      "loss": 0.2003,
+      "step": 2039
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3e-05,
+      "loss": 0.2681,
+      "step": 2040
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3e-05,
+      "loss": 0.2388,
+      "step": 2041
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3e-05,
+      "loss": 0.2006,
+      "step": 2042
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3e-05,
+      "loss": 0.2493,
+      "step": 2043
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3e-05,
+      "loss": 0.2424,
+      "step": 2044
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3e-05,
+      "loss": 0.2329,
+      "step": 2045
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3e-05,
+      "loss": 0.3279,
+      "step": 2046
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3e-05,
+      "loss": 0.2551,
+      "step": 2047
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3e-05,
+      "loss": 0.3706,
+      "step": 2048
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1925,
+      "step": 2049
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3e-05,
+      "loss": 0.2417,
+      "step": 2050
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3e-05,
+      "loss": 0.2089,
+      "step": 2051
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3e-05,
+      "loss": 0.1927,
+      "step": 2052
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3e-05,
+      "loss": 0.1487,
+      "step": 2053
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3e-05,
+      "loss": 0.2859,
+      "step": 2054
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1909,
+      "step": 2055
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3e-05,
+      "loss": 0.4609,
+      "step": 2056
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3e-05,
+      "loss": 0.3279,
+      "step": 2057
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3e-05,
+      "loss": 0.244,
+      "step": 2058
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3e-05,
+      "loss": 0.234,
+      "step": 2059
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3e-05,
+      "loss": 0.2394,
+      "step": 2060
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3e-05,
+      "loss": 0.2078,
+      "step": 2061
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3e-05,
+      "loss": 0.2654,
+      "step": 2062
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3e-05,
+      "loss": 0.2448,
+      "step": 2063
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3e-05,
+      "loss": 0.2152,
+      "step": 2064
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3e-05,
+      "loss": 0.2433,
+      "step": 2065
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3e-05,
+      "loss": 0.2932,
+      "step": 2066
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3e-05,
+      "loss": 0.2012,
+      "step": 2067
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3e-05,
+      "loss": 0.2563,
+      "step": 2068
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3e-05,
+      "loss": 0.163,
+      "step": 2069
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3e-05,
+      "loss": 0.2693,
+      "step": 2070
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3e-05,
+      "loss": 0.312,
+      "step": 2071
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3e-05,
+      "loss": 0.2808,
+      "step": 2072
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3e-05,
+      "loss": 0.1992,
+      "step": 2073
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3e-05,
+      "loss": 0.1095,
+      "step": 2074
+    },
+    {
+      "epoch": 17.0,
+      "eval_loss": 0.294921875,
+      "eval_mae": 0.403564453125,
+      "eval_mse": 0.294921875,
+      "eval_runtime": 51.0904,
+      "eval_samples_per_second": 3237.261,
+      "eval_steps_per_second": 0.274,
+      "step": 2074
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3e-05,
+      "loss": 0.1927,
+      "step": 2075
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3e-05,
+      "loss": 0.2207,
+      "step": 2076
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3e-05,
+      "loss": 0.3196,
+      "step": 2077
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3e-05,
+      "loss": 0.2932,
+      "step": 2078
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3e-05,
+      "loss": 0.3064,
+      "step": 2079
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 3e-05,
+      "loss": 0.245,
+      "step": 2080
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 3e-05,
+      "loss": 0.1544,
+      "step": 2081
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3e-05,
+      "loss": 0.2834,
+      "step": 2082
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3e-05,
+      "loss": 0.207,
+      "step": 2083
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 3e-05,
+      "loss": 0.2744,
+      "step": 2084
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 3e-05,
+      "loss": 0.2068,
+      "step": 2085
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 3e-05,
+      "loss": 0.2637,
+      "step": 2086
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 3e-05,
+      "loss": 0.2505,
+      "step": 2087
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 3e-05,
+      "loss": 0.2827,
+      "step": 2088
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 3e-05,
+      "loss": 0.23,
+      "step": 2089
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 3e-05,
+      "loss": 0.2903,
+      "step": 2090
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 3e-05,
+      "loss": 0.2473,
+      "step": 2091
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 3e-05,
+      "loss": 0.1873,
+      "step": 2092
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 3e-05,
+      "loss": 0.2603,
+      "step": 2093
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 3e-05,
+      "loss": 0.2698,
+      "step": 2094
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 3e-05,
+      "loss": 0.2637,
+      "step": 2095
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 3e-05,
+      "loss": 0.2235,
+      "step": 2096
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 3e-05,
+      "loss": 0.2781,
+      "step": 2097
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3e-05,
+      "loss": 0.2659,
+      "step": 2098
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1862,
+      "step": 2099
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 3e-05,
+      "loss": 0.2766,
+      "step": 2100
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 3e-05,
+      "loss": 0.2279,
+      "step": 2101
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 3e-05,
+      "loss": 0.2462,
+      "step": 2102
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 3e-05,
+      "loss": 0.1879,
+      "step": 2103
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3e-05,
+      "loss": 0.2009,
+      "step": 2104
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3e-05,
+      "loss": 0.2202,
+      "step": 2105
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 3e-05,
+      "loss": 0.2344,
+      "step": 2106
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 3e-05,
+      "loss": 0.2253,
+      "step": 2107
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 3e-05,
+      "loss": 0.2196,
+      "step": 2108
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 3e-05,
+      "loss": 0.171,
+      "step": 2109
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 3e-05,
+      "loss": 0.175,
+      "step": 2110
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 3e-05,
+      "loss": 0.325,
+      "step": 2111
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 3e-05,
+      "loss": 0.1658,
+      "step": 2112
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 3e-05,
+      "loss": 0.162,
+      "step": 2113
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 3e-05,
+      "loss": 0.3064,
+      "step": 2114
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 3e-05,
+      "loss": 0.2761,
+      "step": 2115
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 3e-05,
+      "loss": 0.2542,
+      "step": 2116
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 3e-05,
+      "loss": 0.2781,
+      "step": 2117
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 3e-05,
+      "loss": 0.5747,
+      "step": 2118
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 3e-05,
+      "loss": 0.3638,
+      "step": 2119
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 3e-05,
+      "loss": 0.3242,
+      "step": 2120
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 3e-05,
+      "loss": 0.2285,
+      "step": 2121
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 3e-05,
+      "loss": 0.2336,
+      "step": 2122
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 3e-05,
+      "loss": 0.377,
+      "step": 2123
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 3e-05,
+      "loss": 0.1962,
+      "step": 2124
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 3e-05,
+      "loss": 0.1748,
+      "step": 2125
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 3e-05,
+      "loss": 0.1692,
+      "step": 2126
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 3e-05,
+      "loss": 0.2096,
+      "step": 2127
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 3e-05,
+      "loss": 0.261,
+      "step": 2128
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 3e-05,
+      "loss": 0.2825,
+      "step": 2129
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 3e-05,
+      "loss": 0.2522,
+      "step": 2130
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 3e-05,
+      "loss": 0.1802,
+      "step": 2131
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1497,
+      "step": 2132
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 3e-05,
+      "loss": 0.3352,
+      "step": 2133
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 3e-05,
+      "loss": 0.2333,
+      "step": 2134
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 3e-05,
+      "loss": 0.269,
+      "step": 2135
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 3e-05,
+      "loss": 0.2473,
+      "step": 2136
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1946,
+      "step": 2137
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 3e-05,
+      "loss": 0.213,
+      "step": 2138
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 3e-05,
+      "loss": 0.127,
+      "step": 2139
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 3e-05,
+      "loss": 0.1549,
+      "step": 2140
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 3e-05,
+      "loss": 0.302,
+      "step": 2141
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 3e-05,
+      "loss": 0.2437,
+      "step": 2142
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 3e-05,
+      "loss": 0.238,
+      "step": 2143
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 3e-05,
+      "loss": 0.2372,
+      "step": 2144
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 3e-05,
+      "loss": 0.2856,
+      "step": 2145
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 3e-05,
+      "loss": 0.1295,
+      "step": 2146
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 3e-05,
+      "loss": 0.212,
+      "step": 2147
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 3e-05,
+      "loss": 0.2529,
+      "step": 2148
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 3e-05,
+      "loss": 0.2166,
+      "step": 2149
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 3e-05,
+      "loss": 0.2773,
+      "step": 2150
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 3e-05,
+      "loss": 0.2744,
+      "step": 2151
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 3e-05,
+      "loss": 0.2482,
+      "step": 2152
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 3e-05,
+      "loss": 0.1947,
+      "step": 2153
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 3e-05,
+      "loss": 0.2886,
+      "step": 2154
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 3e-05,
+      "loss": 0.2373,
+      "step": 2155
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 3e-05,
+      "loss": 0.2367,
+      "step": 2156
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 3e-05,
+      "loss": 0.2499,
+      "step": 2157
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 3e-05,
+      "loss": 0.2859,
+      "step": 2158
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 3e-05,
+      "loss": 0.2751,
+      "step": 2159
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 3e-05,
+      "loss": 0.3284,
+      "step": 2160
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 3e-05,
+      "loss": 0.3928,
+      "step": 2161
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 3e-05,
+      "loss": 0.2539,
+      "step": 2162
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 3e-05,
+      "loss": 0.2903,
+      "step": 2163
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 3e-05,
+      "loss": 0.22,
+      "step": 2164
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 3e-05,
+      "loss": 0.2445,
+      "step": 2165
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 3e-05,
+      "loss": 0.272,
+      "step": 2166
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 3e-05,
+      "loss": 0.2271,
+      "step": 2167
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 3e-05,
+      "loss": 0.2429,
+      "step": 2168
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 3e-05,
+      "loss": 0.2346,
+      "step": 2169
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 3e-05,
+      "loss": 0.1765,
+      "step": 2170
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 3e-05,
+      "loss": 0.301,
+      "step": 2171
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 3e-05,
+      "loss": 0.2666,
+      "step": 2172
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 3e-05,
+      "loss": 0.2817,
+      "step": 2173
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 3e-05,
+      "loss": 0.1941,
+      "step": 2174
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 3e-05,
+      "loss": 0.2423,
+      "step": 2175
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 3e-05,
+      "loss": 0.16,
+      "step": 2176
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1769,
+      "step": 2177
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 3e-05,
+      "loss": 0.2705,
+      "step": 2178
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 3e-05,
+      "loss": 0.1702,
+      "step": 2179
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 3e-05,
+      "loss": 0.1667,
+      "step": 2180
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 3e-05,
+      "loss": 0.3721,
+      "step": 2181
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 3e-05,
+      "loss": 0.2505,
+      "step": 2182
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 3e-05,
+      "loss": 0.2605,
+      "step": 2183
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 3e-05,
+      "loss": 0.2524,
+      "step": 2184
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 3e-05,
+      "loss": 0.1692,
+      "step": 2185
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 3e-05,
+      "loss": 0.2206,
+      "step": 2186
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1783,
+      "step": 2187
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 3e-05,
+      "loss": 0.3193,
+      "step": 2188
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 3e-05,
+      "loss": 0.2566,
+      "step": 2189
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 3e-05,
+      "loss": 0.2274,
+      "step": 2190
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 3e-05,
+      "loss": 0.2177,
+      "step": 2191
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 3e-05,
+      "loss": 0.272,
+      "step": 2192
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1697,
+      "step": 2193
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 3e-05,
+      "loss": 0.2878,
+      "step": 2194
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 3e-05,
+      "loss": 0.2715,
+      "step": 2195
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3e-05,
+      "loss": 0.1993,
+      "step": 2196
+    },
+    {
+      "epoch": 18.0,
+      "eval_loss": 0.29443359375,
+      "eval_mae": 0.4033203125,
+      "eval_mse": 0.29443359375,
+      "eval_runtime": 51.3726,
+      "eval_samples_per_second": 3219.476,
+      "eval_steps_per_second": 0.273,
+      "step": 2196
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3e-05,
+      "loss": 0.2615,
+      "step": 2197
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 3e-05,
+      "loss": 0.2556,
+      "step": 2198
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 3e-05,
+      "loss": 0.2869,
+      "step": 2199
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 3e-05,
+      "loss": 0.1786,
+      "step": 2200
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 3e-05,
+      "loss": 0.3303,
+      "step": 2201
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 3e-05,
+      "loss": 0.3682,
+      "step": 2202
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 3e-05,
+      "loss": 0.2595,
+      "step": 2203
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1919,
+      "step": 2204
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 3e-05,
+      "loss": 0.2488,
+      "step": 2205
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 3e-05,
+      "loss": 0.2478,
+      "step": 2206
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 3e-05,
+      "loss": 0.1768,
+      "step": 2207
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 3e-05,
+      "loss": 0.2135,
+      "step": 2208
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 3e-05,
+      "loss": 0.2125,
+      "step": 2209
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1675,
+      "step": 2210
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 3e-05,
+      "loss": 0.1519,
+      "step": 2211
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 3e-05,
+      "loss": 0.1899,
+      "step": 2212
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 3e-05,
+      "loss": 0.2524,
+      "step": 2213
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 3e-05,
+      "loss": 0.1779,
+      "step": 2214
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 3e-05,
+      "loss": 0.155,
+      "step": 2215
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 3e-05,
+      "loss": 0.2189,
+      "step": 2216
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 3e-05,
+      "loss": 0.2245,
+      "step": 2217
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 3e-05,
+      "loss": 0.2856,
+      "step": 2218
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 3e-05,
+      "loss": 0.1898,
+      "step": 2219
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 3e-05,
+      "loss": 0.2764,
+      "step": 2220
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 3e-05,
+      "loss": 0.2084,
+      "step": 2221
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 3e-05,
+      "loss": 0.2954,
+      "step": 2222
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 3e-05,
+      "loss": 0.1908,
+      "step": 2223
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 3e-05,
+      "loss": 0.22,
+      "step": 2224
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 3e-05,
+      "loss": 0.2974,
+      "step": 2225
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 3e-05,
+      "loss": 0.2515,
+      "step": 2226
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 3e-05,
+      "loss": 0.2334,
+      "step": 2227
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 3e-05,
+      "loss": 0.2214,
+      "step": 2228
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 3e-05,
+      "loss": 0.1772,
+      "step": 2229
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 3e-05,
+      "loss": 0.2333,
+      "step": 2230
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 3e-05,
+      "loss": 0.2905,
+      "step": 2231
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 3e-05,
+      "loss": 0.2313,
+      "step": 2232
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1995,
+      "step": 2233
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 3e-05,
+      "loss": 0.2786,
+      "step": 2234
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 3e-05,
+      "loss": 0.2483,
+      "step": 2235
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 3e-05,
+      "loss": 0.2025,
+      "step": 2236
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 3e-05,
+      "loss": 0.2856,
+      "step": 2237
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1996,
+      "step": 2238
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 3e-05,
+      "loss": 0.2507,
+      "step": 2239
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 3e-05,
+      "loss": 0.2524,
+      "step": 2240
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 3e-05,
+      "loss": 0.1461,
+      "step": 2241
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 3e-05,
+      "loss": 0.2429,
+      "step": 2242
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 3e-05,
+      "loss": 0.2302,
+      "step": 2243
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 3e-05,
+      "loss": 0.2025,
+      "step": 2244
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 3e-05,
+      "loss": 0.2471,
+      "step": 2245
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 3e-05,
+      "loss": 0.2832,
+      "step": 2246
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 3e-05,
+      "loss": 0.2603,
+      "step": 2247
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 3e-05,
+      "loss": 0.3093,
+      "step": 2248
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 3e-05,
+      "loss": 0.2522,
+      "step": 2249
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 3e-05,
+      "loss": 0.3538,
+      "step": 2250
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 3e-05,
+      "loss": 0.3137,
+      "step": 2251
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 3e-05,
+      "loss": 0.2366,
+      "step": 2252
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 3e-05,
+      "loss": 0.3481,
+      "step": 2253
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 3e-05,
+      "loss": 0.2289,
+      "step": 2254
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 3e-05,
+      "loss": 0.2126,
+      "step": 2255
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 3e-05,
+      "loss": 0.2405,
+      "step": 2256
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 3e-05,
+      "loss": 0.1637,
+      "step": 2257
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 3e-05,
+      "loss": 0.1661,
+      "step": 2258
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 3e-05,
+      "loss": 0.2072,
+      "step": 2259
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1957,
+      "step": 2260
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 3e-05,
+      "loss": 0.4326,
+      "step": 2261
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 3e-05,
+      "loss": 0.2111,
+      "step": 2262
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 3e-05,
+      "loss": 0.2891,
+      "step": 2263
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 3e-05,
+      "loss": 0.1138,
+      "step": 2264
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 3e-05,
+      "loss": 0.2693,
+      "step": 2265
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 3e-05,
+      "loss": 0.3176,
+      "step": 2266
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 3e-05,
+      "loss": 0.2844,
+      "step": 2267
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 3e-05,
+      "loss": 0.2051,
+      "step": 2268
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 3e-05,
+      "loss": 0.2634,
+      "step": 2269
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 3e-05,
+      "loss": 0.2231,
+      "step": 2270
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 3e-05,
+      "loss": 0.2402,
+      "step": 2271
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 3e-05,
+      "loss": 0.2299,
+      "step": 2272
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 3e-05,
+      "loss": 0.2095,
+      "step": 2273
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 3e-05,
+      "loss": 0.2018,
+      "step": 2274
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 3e-05,
+      "loss": 0.2006,
+      "step": 2275
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 3e-05,
+      "loss": 0.2683,
+      "step": 2276
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 3e-05,
+      "loss": 0.3628,
+      "step": 2277
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 3e-05,
+      "loss": 0.1396,
+      "step": 2278
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 3e-05,
+      "loss": 0.1864,
+      "step": 2279
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 3e-05,
+      "loss": 0.2559,
+      "step": 2280
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 3e-05,
+      "loss": 0.26,
+      "step": 2281
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 3e-05,
+      "loss": 0.2712,
+      "step": 2282
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 3e-05,
+      "loss": 0.2372,
+      "step": 2283
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 3e-05,
+      "loss": 0.2314,
+      "step": 2284
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 3e-05,
+      "loss": 0.2362,
+      "step": 2285
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 3e-05,
+      "loss": 0.1821,
+      "step": 2286
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 3e-05,
+      "loss": 0.2039,
+      "step": 2287
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 3e-05,
+      "loss": 0.1517,
+      "step": 2288
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 3e-05,
+      "loss": 0.4968,
+      "step": 2289
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 3e-05,
+      "loss": 0.2397,
+      "step": 2290
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 3e-05,
+      "loss": 0.1864,
+      "step": 2291
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 3e-05,
+      "loss": 0.229,
+      "step": 2292
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1765,
+      "step": 2293
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 3e-05,
+      "loss": 0.21,
+      "step": 2294
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 3e-05,
+      "loss": 0.2426,
+      "step": 2295
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 3e-05,
+      "loss": 0.2961,
+      "step": 2296
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 3e-05,
+      "loss": 0.2249,
+      "step": 2297
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 3e-05,
+      "loss": 0.4199,
+      "step": 2298
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 3e-05,
+      "loss": 0.2476,
+      "step": 2299
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 3e-05,
+      "loss": 0.2444,
+      "step": 2300
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 3e-05,
+      "loss": 0.2063,
+      "step": 2301
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 3e-05,
+      "loss": 0.2903,
+      "step": 2302
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 3e-05,
+      "loss": 0.2168,
+      "step": 2303
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 3e-05,
+      "loss": 0.27,
+      "step": 2304
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 3e-05,
+      "loss": 0.2583,
+      "step": 2305
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 3e-05,
+      "loss": 0.2202,
+      "step": 2306
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 3e-05,
+      "loss": 0.2595,
+      "step": 2307
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 3e-05,
+      "loss": 0.4358,
+      "step": 2308
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1172,
+      "step": 2309
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 3e-05,
+      "loss": 0.2048,
+      "step": 2310
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 3e-05,
+      "loss": 0.2495,
+      "step": 2311
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 3e-05,
+      "loss": 0.2216,
+      "step": 2312
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 3e-05,
+      "loss": 0.2299,
+      "step": 2313
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 3e-05,
+      "loss": 0.1855,
+      "step": 2314
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 3e-05,
+      "loss": 0.3354,
+      "step": 2315
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 3e-05,
+      "loss": 0.2097,
+      "step": 2316
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 3e-05,
+      "loss": 0.1903,
+      "step": 2317
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 3e-05,
+      "loss": 0.6294,
+      "step": 2318
+    },
+    {
+      "epoch": 19.0,
+      "eval_loss": 0.294677734375,
+      "eval_mae": 0.40234375,
+      "eval_mse": 0.294677734375,
+      "eval_runtime": 51.2332,
+      "eval_samples_per_second": 3228.238,
+      "eval_steps_per_second": 0.273,
+      "step": 2318
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 3e-05,
+      "loss": 0.2524,
+      "step": 2319
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 3e-05,
+      "loss": 0.2761,
+      "step": 2320
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 3e-05,
+      "loss": 0.2891,
+      "step": 2321
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 3e-05,
+      "loss": 0.271,
+      "step": 2322
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 3e-05,
+      "loss": 0.1154,
+      "step": 2323
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 3e-05,
+      "loss": 0.3032,
+      "step": 2324
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 3e-05,
+      "loss": 0.2234,
+      "step": 2325
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1959,
+      "step": 2326
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 3e-05,
+      "loss": 0.2402,
+      "step": 2327
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 3e-05,
+      "loss": 0.2124,
+      "step": 2328
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 3e-05,
+      "loss": 0.1588,
+      "step": 2329
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 3e-05,
+      "loss": 0.2484,
+      "step": 2330
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1971,
+      "step": 2331
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 3e-05,
+      "loss": 0.17,
+      "step": 2332
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 3e-05,
+      "loss": 0.2898,
+      "step": 2333
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 3e-05,
+      "loss": 0.3235,
+      "step": 2334
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 3e-05,
+      "loss": 0.2363,
+      "step": 2335
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 3e-05,
+      "loss": 0.2389,
+      "step": 2336
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 3e-05,
+      "loss": 0.2373,
+      "step": 2337
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1345,
+      "step": 2338
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 3e-05,
+      "loss": 0.1506,
+      "step": 2339
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 3e-05,
+      "loss": 0.3337,
+      "step": 2340
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 3e-05,
+      "loss": 0.1604,
+      "step": 2341
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 3e-05,
+      "loss": 0.2161,
+      "step": 2342
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1733,
+      "step": 2343
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 3e-05,
+      "loss": 0.2588,
+      "step": 2344
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 3e-05,
+      "loss": 0.4033,
+      "step": 2345
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 3e-05,
+      "loss": 0.1274,
+      "step": 2346
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 3e-05,
+      "loss": 0.3374,
+      "step": 2347
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 3e-05,
+      "loss": 0.2087,
+      "step": 2348
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 3e-05,
+      "loss": 0.3425,
+      "step": 2349
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 3e-05,
+      "loss": 0.2363,
+      "step": 2350
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 3e-05,
+      "loss": 0.3374,
+      "step": 2351
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 3e-05,
+      "loss": 0.215,
+      "step": 2352
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 3e-05,
+      "loss": 0.2207,
+      "step": 2353
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1262,
+      "step": 2354
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1726,
+      "step": 2355
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 3e-05,
+      "loss": 0.1859,
+      "step": 2356
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 3e-05,
+      "loss": 0.2881,
+      "step": 2357
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 3e-05,
+      "loss": 0.1635,
+      "step": 2358
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 3e-05,
+      "loss": 0.2184,
+      "step": 2359
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1854,
+      "step": 2360
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 3e-05,
+      "loss": 0.2649,
+      "step": 2361
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 3e-05,
+      "loss": 0.1727,
+      "step": 2362
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 3e-05,
+      "loss": 0.2817,
+      "step": 2363
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 3e-05,
+      "loss": 0.1924,
+      "step": 2364
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1876,
+      "step": 2365
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1497,
+      "step": 2366
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 3e-05,
+      "loss": 0.5957,
+      "step": 2367
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 3e-05,
+      "loss": 0.1559,
+      "step": 2368
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 3e-05,
+      "loss": 0.1412,
+      "step": 2369
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 3e-05,
+      "loss": 0.2445,
+      "step": 2370
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 3e-05,
+      "loss": 0.2332,
+      "step": 2371
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 3e-05,
+      "loss": 0.1497,
+      "step": 2372
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 3e-05,
+      "loss": 0.194,
+      "step": 2373
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 3e-05,
+      "loss": 0.2605,
+      "step": 2374
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 3e-05,
+      "loss": 0.1997,
+      "step": 2375
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1958,
+      "step": 2376
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1277,
+      "step": 2377
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 3e-05,
+      "loss": 0.1945,
+      "step": 2378
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 3e-05,
+      "loss": 0.2222,
+      "step": 2379
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 3e-05,
+      "loss": 0.2632,
+      "step": 2380
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 3e-05,
+      "loss": 0.2269,
+      "step": 2381
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 3e-05,
+      "loss": 0.2224,
+      "step": 2382
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 3e-05,
+      "loss": 0.2277,
+      "step": 2383
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 3e-05,
+      "loss": 0.2664,
+      "step": 2384
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 3e-05,
+      "loss": 0.2456,
+      "step": 2385
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 3e-05,
+      "loss": 0.1511,
+      "step": 2386
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 3e-05,
+      "loss": 0.2179,
+      "step": 2387
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 3e-05,
+      "loss": 0.2494,
+      "step": 2388
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 3e-05,
+      "loss": 0.218,
+      "step": 2389
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 3e-05,
+      "loss": 0.2676,
+      "step": 2390
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 3e-05,
+      "loss": 0.2561,
+      "step": 2391
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 3e-05,
+      "loss": 0.189,
+      "step": 2392
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1859,
+      "step": 2393
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 3e-05,
+      "loss": 0.1952,
+      "step": 2394
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 3e-05,
+      "loss": 0.204,
+      "step": 2395
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 3e-05,
+      "loss": 0.2388,
+      "step": 2396
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 3e-05,
+      "loss": 0.1842,
+      "step": 2397
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1608,
+      "step": 2398
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1594,
+      "step": 2399
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 3e-05,
+      "loss": 0.1709,
+      "step": 2400
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 3e-05,
+      "loss": 0.2109,
+      "step": 2401
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 3e-05,
+      "loss": 0.2876,
+      "step": 2402
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1354,
+      "step": 2403
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 3e-05,
+      "loss": 0.2125,
+      "step": 2404
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 3e-05,
+      "loss": 0.2137,
+      "step": 2405
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 3e-05,
+      "loss": 0.1809,
+      "step": 2406
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 3e-05,
+      "loss": 0.407,
+      "step": 2407
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 3e-05,
+      "loss": 0.1543,
+      "step": 2408
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 3e-05,
+      "loss": 0.222,
+      "step": 2409
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 3e-05,
+      "loss": 0.2383,
+      "step": 2410
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 3e-05,
+      "loss": 0.2455,
+      "step": 2411
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 3e-05,
+      "loss": 0.2925,
+      "step": 2412
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 3e-05,
+      "loss": 0.2532,
+      "step": 2413
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 3e-05,
+      "loss": 0.3569,
+      "step": 2414
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1836,
+      "step": 2415
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 3e-05,
+      "loss": 0.2391,
+      "step": 2416
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 3e-05,
+      "loss": 0.198,
+      "step": 2417
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 3e-05,
+      "loss": 0.2413,
+      "step": 2418
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 3e-05,
+      "loss": 0.2437,
+      "step": 2419
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 3e-05,
+      "loss": 0.2659,
+      "step": 2420
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1973,
+      "step": 2421
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 3e-05,
+      "loss": 0.1945,
+      "step": 2422
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 3e-05,
+      "loss": 0.1356,
+      "step": 2423
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 3e-05,
+      "loss": 0.1975,
+      "step": 2424
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 3e-05,
+      "loss": 0.1716,
+      "step": 2425
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1414,
+      "step": 2426
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 3e-05,
+      "loss": 0.2563,
+      "step": 2427
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 3e-05,
+      "loss": 0.1849,
+      "step": 2428
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 3e-05,
+      "loss": 0.1589,
+      "step": 2429
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 3e-05,
+      "loss": 0.2881,
+      "step": 2430
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 3e-05,
+      "loss": 0.2017,
+      "step": 2431
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1913,
+      "step": 2432
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 3e-05,
+      "loss": 0.2004,
+      "step": 2433
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 3e-05,
+      "loss": 0.1752,
+      "step": 2434
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 3e-05,
+      "loss": 0.1974,
+      "step": 2435
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 3e-05,
+      "loss": 0.2227,
+      "step": 2436
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 3e-05,
+      "loss": 0.2234,
+      "step": 2437
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1638,
+      "step": 2438
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 3e-05,
+      "loss": 0.1403,
+      "step": 2439
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 3e-05,
+      "loss": 0.2681,
+      "step": 2440
+    },
+    {
+      "epoch": 20.0,
+      "eval_loss": 0.282958984375,
+      "eval_mae": 0.393798828125,
+      "eval_mse": 0.282958984375,
+      "eval_runtime": 51.1448,
+      "eval_samples_per_second": 3233.822,
+      "eval_steps_per_second": 0.274,
+      "step": 2440
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 3e-05,
+      "loss": 0.1603,
+      "step": 2441
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 3e-05,
+      "loss": 0.2502,
+      "step": 2442
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 3e-05,
+      "loss": 0.2131,
+      "step": 2443
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 3e-05,
+      "loss": 0.1726,
+      "step": 2444
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 3e-05,
+      "loss": 0.2228,
+      "step": 2445
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 3e-05,
+      "loss": 0.2803,
+      "step": 2446
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 3e-05,
+      "loss": 0.1929,
+      "step": 2447
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1912,
+      "step": 2448
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 3e-05,
+      "loss": 0.3274,
+      "step": 2449
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 3e-05,
+      "loss": 0.124,
+      "step": 2450
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 3e-05,
+      "loss": 0.1388,
+      "step": 2451
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 3e-05,
+      "loss": 0.2605,
+      "step": 2452
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1888,
+      "step": 2453
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 3e-05,
+      "loss": 0.2156,
+      "step": 2454
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 3e-05,
+      "loss": 0.2151,
+      "step": 2455
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 3e-05,
+      "loss": 0.1892,
+      "step": 2456
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 3e-05,
+      "loss": 0.1973,
+      "step": 2457
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 3e-05,
+      "loss": 0.2211,
+      "step": 2458
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 3e-05,
+      "loss": 0.2441,
+      "step": 2459
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1979,
+      "step": 2460
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 3e-05,
+      "loss": 0.1309,
+      "step": 2461
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 3e-05,
+      "loss": 0.2825,
+      "step": 2462
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 3e-05,
+      "loss": 0.1559,
+      "step": 2463
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1416,
+      "step": 2464
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 3e-05,
+      "loss": 0.2834,
+      "step": 2465
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 3e-05,
+      "loss": 0.3196,
+      "step": 2466
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 3e-05,
+      "loss": 0.2537,
+      "step": 2467
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 3e-05,
+      "loss": 0.1748,
+      "step": 2468
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 3e-05,
+      "loss": 0.1528,
+      "step": 2469
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 3e-05,
+      "loss": 0.2474,
+      "step": 2470
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1798,
+      "step": 2471
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 3e-05,
+      "loss": 0.2925,
+      "step": 2472
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 3e-05,
+      "loss": 0.1404,
+      "step": 2473
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 3e-05,
+      "loss": 0.2107,
+      "step": 2474
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 3e-05,
+      "loss": 0.248,
+      "step": 2475
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 3e-05,
+      "loss": 0.2434,
+      "step": 2476
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1842,
+      "step": 2477
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 3e-05,
+      "loss": 0.1925,
+      "step": 2478
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 3e-05,
+      "loss": 0.2047,
+      "step": 2479
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 3e-05,
+      "loss": 0.1713,
+      "step": 2480
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1672,
+      "step": 2481
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 3e-05,
+      "loss": 0.2086,
+      "step": 2482
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 3e-05,
+      "loss": 0.2317,
+      "step": 2483
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 3e-05,
+      "loss": 0.1654,
+      "step": 2484
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 3e-05,
+      "loss": 0.1831,
+      "step": 2485
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 3e-05,
+      "loss": 0.1814,
+      "step": 2486
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 3e-05,
+      "loss": 0.2026,
+      "step": 2487
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1476,
+      "step": 2488
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 3e-05,
+      "loss": 0.3179,
+      "step": 2489
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 3e-05,
+      "loss": 0.1661,
+      "step": 2490
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 3e-05,
+      "loss": 0.2142,
+      "step": 2491
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 3e-05,
+      "loss": 0.1691,
+      "step": 2492
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 3e-05,
+      "loss": 0.2268,
+      "step": 2493
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 3e-05,
+      "loss": 0.2822,
+      "step": 2494
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 3e-05,
+      "loss": 0.186,
+      "step": 2495
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 3e-05,
+      "loss": 0.2529,
+      "step": 2496
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 3e-05,
+      "loss": 0.186,
+      "step": 2497
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1639,
+      "step": 2498
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 3e-05,
+      "loss": 0.2456,
+      "step": 2499
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 3e-05,
+      "loss": 0.1997,
+      "step": 2500
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 3e-05,
+      "loss": 0.2096,
+      "step": 2501
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 3e-05,
+      "loss": 0.2249,
+      "step": 2502
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 3e-05,
+      "loss": 0.2717,
+      "step": 2503
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1456,
+      "step": 2504
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 3e-05,
+      "loss": 0.2181,
+      "step": 2505
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 3e-05,
+      "loss": 0.2393,
+      "step": 2506
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 3e-05,
+      "loss": 0.1462,
+      "step": 2507
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 3e-05,
+      "loss": 0.2225,
+      "step": 2508
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 3e-05,
+      "loss": 0.2208,
+      "step": 2509
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 3e-05,
+      "loss": 0.2644,
+      "step": 2510
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 3e-05,
+      "loss": 0.1689,
+      "step": 2511
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 3e-05,
+      "loss": 0.1538,
+      "step": 2512
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 3e-05,
+      "loss": 0.2686,
+      "step": 2513
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 3e-05,
+      "loss": 0.2808,
+      "step": 2514
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 3e-05,
+      "loss": 0.3003,
+      "step": 2515
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 3e-05,
+      "loss": 0.1995,
+      "step": 2516
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 3e-05,
+      "loss": 0.175,
+      "step": 2517
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 3e-05,
+      "loss": 0.1649,
+      "step": 2518
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 3e-05,
+      "loss": 0.2455,
+      "step": 2519
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1558,
+      "step": 2520
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 3e-05,
+      "loss": 0.2156,
+      "step": 2521
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 3e-05,
+      "loss": 0.1819,
+      "step": 2522
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 3e-05,
+      "loss": 0.2271,
+      "step": 2523
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 3e-05,
+      "loss": 0.1576,
+      "step": 2524
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 3e-05,
+      "loss": 0.246,
+      "step": 2525
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1599,
+      "step": 2526
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 3e-05,
+      "loss": 0.2971,
+      "step": 2527
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 3e-05,
+      "loss": 0.2756,
+      "step": 2528
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 3e-05,
+      "loss": 0.1597,
+      "step": 2529
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 3e-05,
+      "loss": 0.2212,
+      "step": 2530
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 3e-05,
+      "loss": 0.2257,
+      "step": 2531
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 3e-05,
+      "loss": 0.2291,
+      "step": 2532
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 3e-05,
+      "loss": 0.2428,
+      "step": 2533
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 3e-05,
+      "loss": 0.2327,
+      "step": 2534
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 3e-05,
+      "loss": 0.2646,
+      "step": 2535
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 3e-05,
+      "loss": 0.1346,
+      "step": 2536
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1927,
+      "step": 2537
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 3e-05,
+      "loss": 0.2842,
+      "step": 2538
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 3e-05,
+      "loss": 0.3655,
+      "step": 2539
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 3e-05,
+      "loss": 0.2032,
+      "step": 2540
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 3e-05,
+      "loss": 0.2866,
+      "step": 2541
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1605,
+      "step": 2542
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1709,
+      "step": 2543
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 3e-05,
+      "loss": 0.2993,
+      "step": 2544
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 3e-05,
+      "loss": 0.1649,
+      "step": 2545
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 3e-05,
+      "loss": 0.2037,
+      "step": 2546
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 3e-05,
+      "loss": 0.2313,
+      "step": 2547
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1227,
+      "step": 2548
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1858,
+      "step": 2549
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 3e-05,
+      "loss": 0.2766,
+      "step": 2550
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 3e-05,
+      "loss": 0.2534,
+      "step": 2551
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 3e-05,
+      "loss": 0.3025,
+      "step": 2552
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 3e-05,
+      "loss": 0.246,
+      "step": 2553
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 3e-05,
+      "loss": 0.2734,
+      "step": 2554
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 3e-05,
+      "loss": 0.1709,
+      "step": 2555
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 3e-05,
+      "loss": 0.2192,
+      "step": 2556
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 3e-05,
+      "loss": 0.2119,
+      "step": 2557
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 3e-05,
+      "loss": 0.2837,
+      "step": 2558
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 3e-05,
+      "loss": 0.2314,
+      "step": 2559
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 3e-05,
+      "loss": 0.313,
+      "step": 2560
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 3e-05,
+      "loss": 0.2153,
+      "step": 2561
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 3e-05,
+      "loss": 0.4209,
+      "step": 2562
+    },
+    {
+      "epoch": 21.0,
+      "eval_loss": 0.281982421875,
+      "eval_mae": 0.392822265625,
+      "eval_mse": 0.281982421875,
+      "eval_runtime": 51.0711,
+      "eval_samples_per_second": 3238.488,
+      "eval_steps_per_second": 0.274,
+      "step": 2562
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 3e-05,
+      "loss": 0.3108,
+      "step": 2563
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1948,
+      "step": 2564
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1084,
+      "step": 2565
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 3e-05,
+      "loss": 0.2279,
+      "step": 2566
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 3e-05,
+      "loss": 0.1522,
+      "step": 2567
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 3e-05,
+      "loss": 0.1737,
+      "step": 2568
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 3e-05,
+      "loss": 0.2791,
+      "step": 2569
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1296,
+      "step": 2570
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1516,
+      "step": 2571
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 3e-05,
+      "loss": 0.2617,
+      "step": 2572
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 3e-05,
+      "loss": 0.1665,
+      "step": 2573
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 3e-05,
+      "loss": 0.1993,
+      "step": 2574
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 3e-05,
+      "loss": 0.259,
+      "step": 2575
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1825,
+      "step": 2576
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 3e-05,
+      "loss": 0.1877,
+      "step": 2577
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 3e-05,
+      "loss": 0.1844,
+      "step": 2578
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 3e-05,
+      "loss": 0.1805,
+      "step": 2579
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 3e-05,
+      "loss": 0.2073,
+      "step": 2580
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 3e-05,
+      "loss": 0.2632,
+      "step": 2581
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1906,
+      "step": 2582
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 3e-05,
+      "loss": 0.1871,
+      "step": 2583
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 3e-05,
+      "loss": 0.2292,
+      "step": 2584
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 3e-05,
+      "loss": 0.1324,
+      "step": 2585
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 3e-05,
+      "loss": 0.2267,
+      "step": 2586
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 3e-05,
+      "loss": 0.2068,
+      "step": 2587
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 3e-05,
+      "loss": 0.3464,
+      "step": 2588
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 3e-05,
+      "loss": 0.3054,
+      "step": 2589
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 3e-05,
+      "loss": 0.1545,
+      "step": 2590
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 3e-05,
+      "loss": 0.2482,
+      "step": 2591
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 3e-05,
+      "loss": 0.2551,
+      "step": 2592
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 3e-05,
+      "loss": 0.2377,
+      "step": 2593
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 3e-05,
+      "loss": 0.1761,
+      "step": 2594
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 3e-05,
+      "loss": 0.175,
+      "step": 2595
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 3e-05,
+      "loss": 0.1655,
+      "step": 2596
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 3e-05,
+      "loss": 0.2296,
+      "step": 2597
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 3e-05,
+      "loss": 0.2336,
+      "step": 2598
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1637,
+      "step": 2599
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 3e-05,
+      "loss": 0.1981,
+      "step": 2600
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 3e-05,
+      "loss": 0.1541,
+      "step": 2601
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 3e-05,
+      "loss": 0.2009,
+      "step": 2602
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1803,
+      "step": 2603
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1863,
+      "step": 2604
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 3e-05,
+      "loss": 0.2188,
+      "step": 2605
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 3e-05,
+      "loss": 0.21,
+      "step": 2606
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 3e-05,
+      "loss": 0.2065,
+      "step": 2607
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 3e-05,
+      "loss": 0.2727,
+      "step": 2608
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 3e-05,
+      "loss": 0.311,
+      "step": 2609
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 3e-05,
+      "loss": 0.2537,
+      "step": 2610
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 3e-05,
+      "loss": 0.3579,
+      "step": 2611
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 3e-05,
+      "loss": 0.2358,
+      "step": 2612
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 3e-05,
+      "loss": 0.1655,
+      "step": 2613
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 3e-05,
+      "loss": 0.2454,
+      "step": 2614
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 3e-05,
+      "loss": 0.1777,
+      "step": 2615
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 3e-05,
+      "loss": 0.1812,
+      "step": 2616
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 3e-05,
+      "loss": 0.2413,
+      "step": 2617
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 3e-05,
+      "loss": 0.246,
+      "step": 2618
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 3e-05,
+      "loss": 0.199,
+      "step": 2619
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 3e-05,
+      "loss": 0.2426,
+      "step": 2620
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1844,
+      "step": 2621
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 3e-05,
+      "loss": 0.3013,
+      "step": 2622
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 3e-05,
+      "loss": 0.2927,
+      "step": 2623
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 3e-05,
+      "loss": 0.2371,
+      "step": 2624
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 3e-05,
+      "loss": 0.3218,
+      "step": 2625
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 3e-05,
+      "loss": 0.2196,
+      "step": 2626
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 3e-05,
+      "loss": 0.2417,
+      "step": 2627
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 3e-05,
+      "loss": 0.1873,
+      "step": 2628
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 3e-05,
+      "loss": 0.1559,
+      "step": 2629
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 3e-05,
+      "loss": 0.3013,
+      "step": 2630
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 3e-05,
+      "loss": 0.1866,
+      "step": 2631
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 3e-05,
+      "loss": 0.2333,
+      "step": 2632
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 3e-05,
+      "loss": 0.2092,
+      "step": 2633
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 3e-05,
+      "loss": 0.1555,
+      "step": 2634
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 3e-05,
+      "loss": 0.1678,
+      "step": 2635
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 3e-05,
+      "loss": 0.26,
+      "step": 2636
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1954,
+      "step": 2637
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 3e-05,
+      "loss": 0.1852,
+      "step": 2638
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 3e-05,
+      "loss": 0.2534,
+      "step": 2639
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 3e-05,
+      "loss": 0.197,
+      "step": 2640
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 3e-05,
+      "loss": 0.1403,
+      "step": 2641
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 3e-05,
+      "loss": 0.2426,
+      "step": 2642
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1968,
+      "step": 2643
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 3e-05,
+      "loss": 0.1702,
+      "step": 2644
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 3e-05,
+      "loss": 0.1724,
+      "step": 2645
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 3e-05,
+      "loss": 0.2661,
+      "step": 2646
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1992,
+      "step": 2647
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1816,
+      "step": 2648
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 3e-05,
+      "loss": 0.2795,
+      "step": 2649
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 3e-05,
+      "loss": 0.1962,
+      "step": 2650
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 3e-05,
+      "loss": 0.1289,
+      "step": 2651
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 3e-05,
+      "loss": 0.2717,
+      "step": 2652
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 3e-05,
+      "loss": 0.2656,
+      "step": 2653
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 3e-05,
+      "loss": 0.2742,
+      "step": 2654
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 3e-05,
+      "loss": 0.2544,
+      "step": 2655
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 3e-05,
+      "loss": 0.2454,
+      "step": 2656
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 3e-05,
+      "loss": 0.2036,
+      "step": 2657
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 3e-05,
+      "loss": 0.4766,
+      "step": 2658
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 3e-05,
+      "loss": 0.52,
+      "step": 2659
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 3e-05,
+      "loss": 0.2493,
+      "step": 2660
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 3e-05,
+      "loss": 0.4641,
+      "step": 2661
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 3e-05,
+      "loss": 0.2207,
+      "step": 2662
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 3e-05,
+      "loss": 0.1522,
+      "step": 2663
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1805,
+      "step": 2664
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 3e-05,
+      "loss": 0.2179,
+      "step": 2665
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 3e-05,
+      "loss": 0.1572,
+      "step": 2666
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 3e-05,
+      "loss": 0.2249,
+      "step": 2667
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 3e-05,
+      "loss": 0.3513,
+      "step": 2668
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 3e-05,
+      "loss": 0.1842,
+      "step": 2669
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 3e-05,
+      "loss": 0.2086,
+      "step": 2670
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 3e-05,
+      "loss": 0.2151,
+      "step": 2671
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 3e-05,
+      "loss": 0.1272,
+      "step": 2672
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 3e-05,
+      "loss": 0.2325,
+      "step": 2673
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 3e-05,
+      "loss": 0.2263,
+      "step": 2674
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1748,
+      "step": 2675
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1953,
+      "step": 2676
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 3e-05,
+      "loss": 0.1954,
+      "step": 2677
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 3e-05,
+      "loss": 0.249,
+      "step": 2678
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 3e-05,
+      "loss": 0.2277,
+      "step": 2679
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 3e-05,
+      "loss": 0.2119,
+      "step": 2680
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 3e-05,
+      "loss": 0.2499,
+      "step": 2681
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 3e-05,
+      "loss": 0.2198,
+      "step": 2682
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 3e-05,
+      "loss": 0.1545,
+      "step": 2683
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 3e-05,
+      "loss": 0.3176,
+      "step": 2684
+    },
+    {
+      "epoch": 22.0,
+      "eval_loss": 0.2783203125,
+      "eval_mae": 0.389892578125,
+      "eval_mse": 0.2783203125,
+      "eval_runtime": 51.1527,
+      "eval_samples_per_second": 3233.317,
+      "eval_steps_per_second": 0.274,
+      "step": 2684
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 3e-05,
+      "loss": 0.23,
+      "step": 2685
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 3e-05,
+      "loss": 0.2085,
+      "step": 2686
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1536,
+      "step": 2687
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 3e-05,
+      "loss": 0.1523,
+      "step": 2688
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 3e-05,
+      "loss": 0.1741,
+      "step": 2689
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 3e-05,
+      "loss": 0.1326,
+      "step": 2690
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 3e-05,
+      "loss": 0.2445,
+      "step": 2691
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 3e-05,
+      "loss": 0.2598,
+      "step": 2692
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 3e-05,
+      "loss": 0.2142,
+      "step": 2693
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 3e-05,
+      "loss": 0.2192,
+      "step": 2694
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 3e-05,
+      "loss": 0.1982,
+      "step": 2695
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 3e-05,
+      "loss": 0.1925,
+      "step": 2696
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1519,
+      "step": 2697
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 3e-05,
+      "loss": 0.2272,
+      "step": 2698
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 3e-05,
+      "loss": 0.1954,
+      "step": 2699
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 3e-05,
+      "loss": 0.1656,
+      "step": 2700
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 3e-05,
+      "loss": 0.1868,
+      "step": 2701
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 3e-05,
+      "loss": 0.1771,
+      "step": 2702
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1716,
+      "step": 2703
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1313,
+      "step": 2704
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 3e-05,
+      "loss": 0.1904,
+      "step": 2705
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 3e-05,
+      "loss": 0.2043,
+      "step": 2706
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 3e-05,
+      "loss": 0.2163,
+      "step": 2707
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1981,
+      "step": 2708
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1873,
+      "step": 2709
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 3e-05,
+      "loss": 0.2002,
+      "step": 2710
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 3e-05,
+      "loss": 0.2434,
+      "step": 2711
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 3e-05,
+      "loss": 0.1595,
+      "step": 2712
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 3e-05,
+      "loss": 0.1476,
+      "step": 2713
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1984,
+      "step": 2714
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 3e-05,
+      "loss": 0.2067,
+      "step": 2715
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 3e-05,
+      "loss": 0.1947,
+      "step": 2716
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 3e-05,
+      "loss": 0.2534,
+      "step": 2717
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 3e-05,
+      "loss": 0.2203,
+      "step": 2718
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 3e-05,
+      "loss": 0.1754,
+      "step": 2719
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 3e-05,
+      "loss": 0.2051,
+      "step": 2720
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 3e-05,
+      "loss": 0.2062,
+      "step": 2721
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 3e-05,
+      "loss": 0.208,
+      "step": 2722
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 3e-05,
+      "loss": 0.1978,
+      "step": 2723
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 3e-05,
+      "loss": 0.2097,
+      "step": 2724
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 3e-05,
+      "loss": 0.26,
+      "step": 2725
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 3e-05,
+      "loss": 0.2499,
+      "step": 2726
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 3e-05,
+      "loss": 0.1519,
+      "step": 2727
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 3e-05,
+      "loss": 0.2935,
+      "step": 2728
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 3e-05,
+      "loss": 0.2817,
+      "step": 2729
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 3e-05,
+      "loss": 0.3118,
+      "step": 2730
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 3e-05,
+      "loss": 0.3044,
+      "step": 2731
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1898,
+      "step": 2732
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 3e-05,
+      "loss": 0.1812,
+      "step": 2733
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 3e-05,
+      "loss": 0.1317,
+      "step": 2734
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 3e-05,
+      "loss": 0.1508,
+      "step": 2735
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 3e-05,
+      "loss": 0.1927,
+      "step": 2736
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 3e-05,
+      "loss": 0.2915,
+      "step": 2737
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 3e-05,
+      "loss": 0.1642,
+      "step": 2738
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 3e-05,
+      "loss": 0.178,
+      "step": 2739
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 3e-05,
+      "loss": 0.2372,
+      "step": 2740
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 3e-05,
+      "loss": 0.3772,
+      "step": 2741
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1572,
+      "step": 2742
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 3e-05,
+      "loss": 0.2206,
+      "step": 2743
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 3e-05,
+      "loss": 0.2079,
+      "step": 2744
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 3e-05,
+      "loss": 0.199,
+      "step": 2745
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 3e-05,
+      "loss": 0.1143,
+      "step": 2746
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 3e-05,
+      "loss": 0.283,
+      "step": 2747
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1807,
+      "step": 2748
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 3e-05,
+      "loss": 0.1479,
+      "step": 2749
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 3e-05,
+      "loss": 0.2168,
+      "step": 2750
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 3e-05,
+      "loss": 0.121,
+      "step": 2751
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 3e-05,
+      "loss": 0.1764,
+      "step": 2752
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 3e-05,
+      "loss": 0.1898,
+      "step": 2753
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 3e-05,
+      "loss": 0.1986,
+      "step": 2754
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 3e-05,
+      "loss": 0.2341,
+      "step": 2755
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 3e-05,
+      "loss": 0.2179,
+      "step": 2756
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 3e-05,
+      "loss": 0.2332,
+      "step": 2757
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1519,
+      "step": 2758
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 3e-05,
+      "loss": 0.302,
+      "step": 2759
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 3e-05,
+      "loss": 0.219,
+      "step": 2760
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 3e-05,
+      "loss": 0.2346,
+      "step": 2761
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 3e-05,
+      "loss": 0.2438,
+      "step": 2762
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 3e-05,
+      "loss": 0.2664,
+      "step": 2763
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1746,
+      "step": 2764
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1877,
+      "step": 2765
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 3e-05,
+      "loss": 0.1837,
+      "step": 2766
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 3e-05,
+      "loss": 0.2854,
+      "step": 2767
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 3e-05,
+      "loss": 0.1774,
+      "step": 2768
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 3e-05,
+      "loss": 0.2517,
+      "step": 2769
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1857,
+      "step": 2770
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 3e-05,
+      "loss": 0.139,
+      "step": 2771
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 3e-05,
+      "loss": 0.2115,
+      "step": 2772
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 3e-05,
+      "loss": 0.158,
+      "step": 2773
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 3e-05,
+      "loss": 0.2311,
+      "step": 2774
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 3e-05,
+      "loss": 0.2338,
+      "step": 2775
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 3e-05,
+      "loss": 0.1503,
+      "step": 2776
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 3e-05,
+      "loss": 0.2396,
+      "step": 2777
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 3e-05,
+      "loss": 0.225,
+      "step": 2778
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 3e-05,
+      "loss": 0.2305,
+      "step": 2779
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 3e-05,
+      "loss": 0.2722,
+      "step": 2780
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1332,
+      "step": 2781
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1804,
+      "step": 2782
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 3e-05,
+      "loss": 0.2067,
+      "step": 2783
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 3e-05,
+      "loss": 0.1987,
+      "step": 2784
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 3e-05,
+      "loss": 0.2969,
+      "step": 2785
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1846,
+      "step": 2786
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1868,
+      "step": 2787
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 3e-05,
+      "loss": 0.2024,
+      "step": 2788
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 3e-05,
+      "loss": 0.2961,
+      "step": 2789
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 3e-05,
+      "loss": 0.1968,
+      "step": 2790
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 3e-05,
+      "loss": 0.1849,
+      "step": 2791
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1947,
+      "step": 2792
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 3e-05,
+      "loss": 0.2185,
+      "step": 2793
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 3e-05,
+      "loss": 0.201,
+      "step": 2794
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 3e-05,
+      "loss": 0.1637,
+      "step": 2795
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 3e-05,
+      "loss": 0.1172,
+      "step": 2796
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 3e-05,
+      "loss": 0.2303,
+      "step": 2797
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 3e-05,
+      "loss": 0.177,
+      "step": 2798
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 3e-05,
+      "loss": 0.2869,
+      "step": 2799
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 3e-05,
+      "loss": 0.1866,
+      "step": 2800
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 3e-05,
+      "loss": 0.1367,
+      "step": 2801
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 3e-05,
+      "loss": 0.3608,
+      "step": 2802
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1736,
+      "step": 2803
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 3e-05,
+      "loss": 0.152,
+      "step": 2804
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 3e-05,
+      "loss": 0.2761,
+      "step": 2805
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 3e-05,
+      "loss": 0.1823,
+      "step": 2806
+    },
+    {
+      "epoch": 23.0,
+      "eval_loss": 0.2802734375,
+      "eval_mae": 0.39013671875,
+      "eval_mse": 0.2802734375,
+      "eval_runtime": 51.0642,
+      "eval_samples_per_second": 3238.922,
+      "eval_steps_per_second": 0.274,
+      "step": 2806
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 3e-05,
+      "loss": 0.3398,
+      "step": 2807
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1477,
+      "step": 2808
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 3e-05,
+      "loss": 0.2935,
+      "step": 2809
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 3e-05,
+      "loss": 0.1825,
+      "step": 2810
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 3e-05,
+      "loss": 0.2656,
+      "step": 2811
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 3e-05,
+      "loss": 0.2224,
+      "step": 2812
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 3e-05,
+      "loss": 0.241,
+      "step": 2813
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 3e-05,
+      "loss": 0.2576,
+      "step": 2814
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1175,
+      "step": 2815
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 3e-05,
+      "loss": 0.2917,
+      "step": 2816
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 3e-05,
+      "loss": 0.1729,
+      "step": 2817
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 3e-05,
+      "loss": 0.2693,
+      "step": 2818
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 3e-05,
+      "loss": 0.2015,
+      "step": 2819
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1477,
+      "step": 2820
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 3e-05,
+      "loss": 0.1748,
+      "step": 2821
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 3e-05,
+      "loss": 0.2625,
+      "step": 2822
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 3e-05,
+      "loss": 0.312,
+      "step": 2823
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 3e-05,
+      "loss": 0.2437,
+      "step": 2824
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 3e-05,
+      "loss": 0.2434,
+      "step": 2825
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 3e-05,
+      "loss": 0.193,
+      "step": 2826
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 3e-05,
+      "loss": 0.2267,
+      "step": 2827
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 3e-05,
+      "loss": 0.3301,
+      "step": 2828
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 3e-05,
+      "loss": 0.2754,
+      "step": 2829
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 3e-05,
+      "loss": 0.2112,
+      "step": 2830
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1716,
+      "step": 2831
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 3e-05,
+      "loss": 0.0962,
+      "step": 2832
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 3e-05,
+      "loss": 0.1815,
+      "step": 2833
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 3e-05,
+      "loss": 0.2452,
+      "step": 2834
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 3e-05,
+      "loss": 0.1536,
+      "step": 2835
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1639,
+      "step": 2836
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 3e-05,
+      "loss": 0.2529,
+      "step": 2837
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 3e-05,
+      "loss": 0.1877,
+      "step": 2838
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 3e-05,
+      "loss": 0.1478,
+      "step": 2839
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 3e-05,
+      "loss": 0.2406,
+      "step": 2840
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 3e-05,
+      "loss": 0.2937,
+      "step": 2841
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 3e-05,
+      "loss": 0.2101,
+      "step": 2842
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 3e-05,
+      "loss": 0.2849,
+      "step": 2843
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 3e-05,
+      "loss": 0.2411,
+      "step": 2844
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 3e-05,
+      "loss": 0.2915,
+      "step": 2845
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 3e-05,
+      "loss": 0.1273,
+      "step": 2846
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1736,
+      "step": 2847
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 3e-05,
+      "loss": 0.166,
+      "step": 2848
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 3e-05,
+      "loss": 0.165,
+      "step": 2849
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 3e-05,
+      "loss": 0.1803,
+      "step": 2850
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 3e-05,
+      "loss": 0.2069,
+      "step": 2851
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 3e-05,
+      "loss": 0.1882,
+      "step": 2852
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 3e-05,
+      "loss": 0.2654,
+      "step": 2853
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1646,
+      "step": 2854
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 3e-05,
+      "loss": 0.113,
+      "step": 2855
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 3e-05,
+      "loss": 0.2212,
+      "step": 2856
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 3e-05,
+      "loss": 0.166,
+      "step": 2857
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 3e-05,
+      "loss": 0.1527,
+      "step": 2858
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 3e-05,
+      "loss": 0.1798,
+      "step": 2859
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 3e-05,
+      "loss": 0.2046,
+      "step": 2860
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 3e-05,
+      "loss": 0.3862,
+      "step": 2861
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 3e-05,
+      "loss": 0.2087,
+      "step": 2862
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 3e-05,
+      "loss": 0.2083,
+      "step": 2863
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1488,
+      "step": 2864
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 3e-05,
+      "loss": 0.2249,
+      "step": 2865
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 3e-05,
+      "loss": 0.2245,
+      "step": 2866
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 3e-05,
+      "loss": 0.1604,
+      "step": 2867
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 3e-05,
+      "loss": 0.1854,
+      "step": 2868
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1339,
+      "step": 2869
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1599,
+      "step": 2870
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 3e-05,
+      "loss": 0.1677,
+      "step": 2871
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 3e-05,
+      "loss": 0.177,
+      "step": 2872
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 3e-05,
+      "loss": 0.1715,
+      "step": 2873
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 3e-05,
+      "loss": 0.1825,
+      "step": 2874
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 3e-05,
+      "loss": 0.2031,
+      "step": 2875
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 3e-05,
+      "loss": 0.2383,
+      "step": 2876
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 3e-05,
+      "loss": 0.249,
+      "step": 2877
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 3e-05,
+      "loss": 0.1768,
+      "step": 2878
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 3e-05,
+      "loss": 0.1984,
+      "step": 2879
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 3e-05,
+      "loss": 0.2622,
+      "step": 2880
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 3e-05,
+      "loss": 0.192,
+      "step": 2881
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 3e-05,
+      "loss": 0.1536,
+      "step": 2882
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 3e-05,
+      "loss": 0.1388,
+      "step": 2883
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 3e-05,
+      "loss": 0.2042,
+      "step": 2884
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 3e-05,
+      "loss": 0.218,
+      "step": 2885
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1288,
+      "step": 2886
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 3e-05,
+      "loss": 0.2595,
+      "step": 2887
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 3e-05,
+      "loss": 0.1877,
+      "step": 2888
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 3e-05,
+      "loss": 0.1638,
+      "step": 2889
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 3e-05,
+      "loss": 0.2014,
+      "step": 2890
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1401,
+      "step": 2891
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 3e-05,
+      "loss": 0.224,
+      "step": 2892
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 3e-05,
+      "loss": 0.2639,
+      "step": 2893
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 3e-05,
+      "loss": 0.166,
+      "step": 2894
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 3e-05,
+      "loss": 0.17,
+      "step": 2895
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 3e-05,
+      "loss": 0.2546,
+      "step": 2896
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 3e-05,
+      "loss": 0.1666,
+      "step": 2897
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 3e-05,
+      "loss": 0.1387,
+      "step": 2898
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 3e-05,
+      "loss": 0.2063,
+      "step": 2899
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 3e-05,
+      "loss": 0.224,
+      "step": 2900
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 3e-05,
+      "loss": 0.2166,
+      "step": 2901
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 3e-05,
+      "loss": 0.2539,
+      "step": 2902
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1478,
+      "step": 2903
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1707,
+      "step": 2904
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 3e-05,
+      "loss": 0.2108,
+      "step": 2905
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 3e-05,
+      "loss": 0.2272,
+      "step": 2906
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 3e-05,
+      "loss": 0.1686,
+      "step": 2907
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1948,
+      "step": 2908
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 3e-05,
+      "loss": 0.2891,
+      "step": 2909
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 3e-05,
+      "loss": 0.1863,
+      "step": 2910
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 3e-05,
+      "loss": 0.2211,
+      "step": 2911
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 3e-05,
+      "loss": 0.291,
+      "step": 2912
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 3e-05,
+      "loss": 0.1747,
+      "step": 2913
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 3e-05,
+      "loss": 0.3596,
+      "step": 2914
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 3e-05,
+      "loss": 0.2327,
+      "step": 2915
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 3e-05,
+      "loss": 0.1974,
+      "step": 2916
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 3e-05,
+      "loss": 0.1948,
+      "step": 2917
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 3e-05,
+      "loss": 0.1591,
+      "step": 2918
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 3e-05,
+      "loss": 0.2517,
+      "step": 2919
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1346,
+      "step": 2920
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 3e-05,
+      "loss": 0.2329,
+      "step": 2921
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 3e-05,
+      "loss": 0.1951,
+      "step": 2922
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 3e-05,
+      "loss": 0.1765,
+      "step": 2923
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 3e-05,
+      "loss": 0.1456,
+      "step": 2924
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 3e-05,
+      "loss": 0.3503,
+      "step": 2925
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 3e-05,
+      "loss": 0.155,
+      "step": 2926
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 3e-05,
+      "loss": 0.2113,
+      "step": 2927
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 3e-05,
+      "loss": 0.1149,
+      "step": 2928
+    },
+    {
+      "epoch": 24.0,
+      "eval_loss": 0.27685546875,
+      "eval_mae": 0.38671875,
+      "eval_mse": 0.27685546875,
+      "eval_runtime": 51.1303,
+      "eval_samples_per_second": 3234.734,
+      "eval_steps_per_second": 0.274,
+      "step": 2928
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 3e-05,
+      "loss": 0.2068,
+      "step": 2929
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1704,
+      "step": 2930
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 3e-05,
+      "loss": 0.2617,
+      "step": 2931
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 3e-05,
+      "loss": 0.1375,
+      "step": 2932
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 3e-05,
+      "loss": 0.2437,
+      "step": 2933
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 3e-05,
+      "loss": 0.1702,
+      "step": 2934
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 3e-05,
+      "loss": 0.1168,
+      "step": 2935
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1284,
+      "step": 2936
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1952,
+      "step": 2937
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 3e-05,
+      "loss": 0.1807,
+      "step": 2938
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 3e-05,
+      "loss": 0.1915,
+      "step": 2939
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 3e-05,
+      "loss": 0.1401,
+      "step": 2940
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 3e-05,
+      "loss": 0.2218,
+      "step": 2941
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1674,
+      "step": 2942
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 3e-05,
+      "loss": 0.2195,
+      "step": 2943
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 3e-05,
+      "loss": 0.313,
+      "step": 2944
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 3e-05,
+      "loss": 0.1759,
+      "step": 2945
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 3e-05,
+      "loss": 0.2861,
+      "step": 2946
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1251,
+      "step": 2947
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1533,
+      "step": 2948
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 3e-05,
+      "loss": 0.2527,
+      "step": 2949
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 3e-05,
+      "loss": 0.125,
+      "step": 2950
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 3e-05,
+      "loss": 0.1802,
+      "step": 2951
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1522,
+      "step": 2952
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1991,
+      "step": 2953
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 3e-05,
+      "loss": 0.156,
+      "step": 2954
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 3e-05,
+      "loss": 0.136,
+      "step": 2955
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 3e-05,
+      "loss": 0.1454,
+      "step": 2956
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 3e-05,
+      "loss": 0.2101,
+      "step": 2957
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 3e-05,
+      "loss": 0.215,
+      "step": 2958
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1087,
+      "step": 2959
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 3e-05,
+      "loss": 0.2026,
+      "step": 2960
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 3e-05,
+      "loss": 0.1372,
+      "step": 2961
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 3e-05,
+      "loss": 0.1932,
+      "step": 2962
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 3e-05,
+      "loss": 0.1327,
+      "step": 2963
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1729,
+      "step": 2964
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1936,
+      "step": 2965
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 3e-05,
+      "loss": 0.1406,
+      "step": 2966
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 3e-05,
+      "loss": 0.2347,
+      "step": 2967
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 3e-05,
+      "loss": 0.2065,
+      "step": 2968
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 3e-05,
+      "loss": 0.2144,
+      "step": 2969
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1849,
+      "step": 2970
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 3e-05,
+      "loss": 0.2307,
+      "step": 2971
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 3e-05,
+      "loss": 0.2155,
+      "step": 2972
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 3e-05,
+      "loss": 0.2456,
+      "step": 2973
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 3e-05,
+      "loss": 0.1976,
+      "step": 2974
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1718,
+      "step": 2975
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 3e-05,
+      "loss": 0.2546,
+      "step": 2976
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 3e-05,
+      "loss": 0.2159,
+      "step": 2977
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 3e-05,
+      "loss": 0.1345,
+      "step": 2978
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 3e-05,
+      "loss": 0.1824,
+      "step": 2979
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 3e-05,
+      "loss": 0.2083,
+      "step": 2980
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 3e-05,
+      "loss": 0.2289,
+      "step": 2981
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 3e-05,
+      "loss": 0.1375,
+      "step": 2982
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 3e-05,
+      "loss": 0.2239,
+      "step": 2983
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 3e-05,
+      "loss": 0.2336,
+      "step": 2984
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 3e-05,
+      "loss": 0.1459,
+      "step": 2985
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1555,
+      "step": 2986
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 3e-05,
+      "loss": 0.2213,
+      "step": 2987
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 3e-05,
+      "loss": 0.1592,
+      "step": 2988
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 3e-05,
+      "loss": 0.118,
+      "step": 2989
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 3e-05,
+      "loss": 0.1281,
+      "step": 2990
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1487,
+      "step": 2991
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 3e-05,
+      "loss": 0.228,
+      "step": 2992
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 3e-05,
+      "loss": 0.2202,
+      "step": 2993
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 3e-05,
+      "loss": 0.1848,
+      "step": 2994
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 3e-05,
+      "loss": 0.1384,
+      "step": 2995
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 3e-05,
+      "loss": 0.1368,
+      "step": 2996
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 3e-05,
+      "loss": 0.2252,
+      "step": 2997
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 3e-05,
+      "loss": 0.191,
+      "step": 2998
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 3e-05,
+      "loss": 0.2061,
+      "step": 2999
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 3e-05,
+      "loss": 0.2102,
+      "step": 3000
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 3e-05,
+      "loss": 0.1561,
+      "step": 3001
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 3e-05,
+      "loss": 0.2175,
+      "step": 3002
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 3e-05,
+      "loss": 0.184,
+      "step": 3003
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 3e-05,
+      "loss": 0.1815,
+      "step": 3004
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 3e-05,
+      "loss": 0.2101,
+      "step": 3005
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 3e-05,
+      "loss": 0.1335,
+      "step": 3006
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 3e-05,
+      "loss": 0.1726,
+      "step": 3007
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 3e-05,
+      "loss": 0.2664,
+      "step": 3008
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 3e-05,
+      "loss": 0.2443,
+      "step": 3009
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 3e-05,
+      "loss": 0.2123,
+      "step": 3010
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 3e-05,
+      "loss": 0.2666,
+      "step": 3011
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 3e-05,
+      "loss": 0.1619,
+      "step": 3012
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1354,
+      "step": 3013
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1287,
+      "step": 3014
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 3e-05,
+      "loss": 0.2234,
+      "step": 3015
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 3e-05,
+      "loss": 0.2401,
+      "step": 3016
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 3e-05,
+      "loss": 0.1729,
+      "step": 3017
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 3e-05,
+      "loss": 0.1968,
+      "step": 3018
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 3e-05,
+      "loss": 0.2737,
+      "step": 3019
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 3e-05,
+      "loss": 0.162,
+      "step": 3020
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 3e-05,
+      "loss": 0.1678,
+      "step": 3021
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 3e-05,
+      "loss": 0.1531,
+      "step": 3022
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 3e-05,
+      "loss": 0.2332,
+      "step": 3023
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 3e-05,
+      "loss": 0.2212,
+      "step": 3024
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1888,
+      "step": 3025
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1697,
+      "step": 3026
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 3e-05,
+      "loss": 0.1503,
+      "step": 3027
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 3e-05,
+      "loss": 0.2043,
+      "step": 3028
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 3e-05,
+      "loss": 0.1725,
+      "step": 3029
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 3e-05,
+      "loss": 0.2252,
+      "step": 3030
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 3e-05,
+      "loss": 0.219,
+      "step": 3031
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 3e-05,
+      "loss": 0.189,
+      "step": 3032
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 3e-05,
+      "loss": 0.1318,
+      "step": 3033
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 3e-05,
+      "loss": 0.2482,
+      "step": 3034
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 3e-05,
+      "loss": 0.1951,
+      "step": 3035
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 3e-05,
+      "loss": 0.2656,
+      "step": 3036
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 3e-05,
+      "loss": 0.2031,
+      "step": 3037
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 3e-05,
+      "loss": 0.3494,
+      "step": 3038
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 3e-05,
+      "loss": 0.1393,
+      "step": 3039
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 3e-05,
+      "loss": 0.1719,
+      "step": 3040
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 3e-05,
+      "loss": 0.2153,
+      "step": 3041
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1611,
+      "step": 3042
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 3e-05,
+      "loss": 0.246,
+      "step": 3043
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 3e-05,
+      "loss": 0.1691,
+      "step": 3044
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 3e-05,
+      "loss": 0.2627,
+      "step": 3045
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 3e-05,
+      "loss": 0.1659,
+      "step": 3046
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1118,
+      "step": 3047
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1593,
+      "step": 3048
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 3e-05,
+      "loss": 0.1796,
+      "step": 3049
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 3e-05,
+      "loss": 0.2576,
+      "step": 3050
+    },
+    {
+      "epoch": 25.0,
+      "eval_loss": 0.276611328125,
+      "eval_mae": 0.38720703125,
+      "eval_mse": 0.276611328125,
+      "eval_runtime": 51.084,
+      "eval_samples_per_second": 3237.666,
+      "eval_steps_per_second": 0.274,
+      "step": 3050
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 3e-05,
+      "loss": 0.2452,
+      "step": 3051
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1477,
+      "step": 3052
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1947,
+      "step": 3053
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 3e-05,
+      "loss": 0.1644,
+      "step": 3054
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 3e-05,
+      "loss": 0.3032,
+      "step": 3055
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 3e-05,
+      "loss": 0.1494,
+      "step": 3056
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 3e-05,
+      "loss": 0.1934,
+      "step": 3057
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 3e-05,
+      "loss": 0.3582,
+      "step": 3058
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1899,
+      "step": 3059
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 3e-05,
+      "loss": 0.1549,
+      "step": 3060
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 3e-05,
+      "loss": 0.1635,
+      "step": 3061
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 3e-05,
+      "loss": 0.1445,
+      "step": 3062
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 3e-05,
+      "loss": 0.127,
+      "step": 3063
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 3e-05,
+      "loss": 0.2194,
+      "step": 3064
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 3e-05,
+      "loss": 0.17,
+      "step": 3065
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 3e-05,
+      "loss": 0.1681,
+      "step": 3066
+    },
+    {
+      "epoch": 25.14,
+      "learning_rate": 3e-05,
+      "loss": 0.1216,
+      "step": 3067
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 3e-05,
+      "loss": 0.183,
+      "step": 3068
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1444,
+      "step": 3069
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 3e-05,
+      "loss": 0.2959,
+      "step": 3070
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 3e-05,
+      "loss": 0.1781,
+      "step": 3071
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 3e-05,
+      "loss": 0.2365,
+      "step": 3072
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 3e-05,
+      "loss": 0.1451,
+      "step": 3073
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1995,
+      "step": 3074
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1265,
+      "step": 3075
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 3e-05,
+      "loss": 0.2375,
+      "step": 3076
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 3e-05,
+      "loss": 0.171,
+      "step": 3077
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 3e-05,
+      "loss": 0.2532,
+      "step": 3078
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 3e-05,
+      "loss": 0.1738,
+      "step": 3079
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 3e-05,
+      "loss": 0.2502,
+      "step": 3080
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 3e-05,
+      "loss": 0.2178,
+      "step": 3081
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 3e-05,
+      "loss": 0.1718,
+      "step": 3082
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 3e-05,
+      "loss": 0.2993,
+      "step": 3083
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 3e-05,
+      "loss": 0.1664,
+      "step": 3084
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 3e-05,
+      "loss": 0.1979,
+      "step": 3085
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1442,
+      "step": 3086
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1815,
+      "step": 3087
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 3e-05,
+      "loss": 0.16,
+      "step": 3088
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 3e-05,
+      "loss": 0.2074,
+      "step": 3089
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 3e-05,
+      "loss": 0.1522,
+      "step": 3090
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1669,
+      "step": 3091
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 3e-05,
+      "loss": 0.2399,
+      "step": 3092
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 3e-05,
+      "loss": 0.1992,
+      "step": 3093
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 3e-05,
+      "loss": 0.1659,
+      "step": 3094
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 3e-05,
+      "loss": 0.1282,
+      "step": 3095
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 3e-05,
+      "loss": 0.3845,
+      "step": 3096
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 3e-05,
+      "loss": 0.0902,
+      "step": 3097
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1361,
+      "step": 3098
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 3e-05,
+      "loss": 0.2676,
+      "step": 3099
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 3e-05,
+      "loss": 0.3215,
+      "step": 3100
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 3e-05,
+      "loss": 0.1785,
+      "step": 3101
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 3e-05,
+      "loss": 0.2964,
+      "step": 3102
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 3e-05,
+      "loss": 0.2318,
+      "step": 3103
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 3e-05,
+      "loss": 0.3301,
+      "step": 3104
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 3e-05,
+      "loss": 0.2446,
+      "step": 3105
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 3e-05,
+      "loss": 0.1964,
+      "step": 3106
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 3e-05,
+      "loss": 0.2202,
+      "step": 3107
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1015,
+      "step": 3108
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 3e-05,
+      "loss": 0.2012,
+      "step": 3109
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 3e-05,
+      "loss": 0.2483,
+      "step": 3110
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 3e-05,
+      "loss": 0.131,
+      "step": 3111
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 3e-05,
+      "loss": 0.1788,
+      "step": 3112
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1515,
+      "step": 3113
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 3e-05,
+      "loss": 0.2234,
+      "step": 3114
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 3e-05,
+      "loss": 0.1661,
+      "step": 3115
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 3e-05,
+      "loss": 0.1503,
+      "step": 3116
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 3e-05,
+      "loss": 0.2374,
+      "step": 3117
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 3e-05,
+      "loss": 0.1953,
+      "step": 3118
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 3e-05,
+      "loss": 0.2168,
+      "step": 3119
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 3e-05,
+      "loss": 0.2166,
+      "step": 3120
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 3e-05,
+      "loss": 0.1598,
+      "step": 3121
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 3e-05,
+      "loss": 0.1647,
+      "step": 3122
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 3e-05,
+      "loss": 0.1473,
+      "step": 3123
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1241,
+      "step": 3124
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 3e-05,
+      "loss": 0.123,
+      "step": 3125
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 3e-05,
+      "loss": 0.1416,
+      "step": 3126
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 3e-05,
+      "loss": 0.175,
+      "step": 3127
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 3e-05,
+      "loss": 0.1566,
+      "step": 3128
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 3e-05,
+      "loss": 0.3208,
+      "step": 3129
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 3e-05,
+      "loss": 0.158,
+      "step": 3130
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1915,
+      "step": 3131
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 3e-05,
+      "loss": 0.2424,
+      "step": 3132
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 3e-05,
+      "loss": 0.2172,
+      "step": 3133
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 3e-05,
+      "loss": 0.2258,
+      "step": 3134
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 3e-05,
+      "loss": 0.2751,
+      "step": 3135
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1638,
+      "step": 3136
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 3e-05,
+      "loss": 0.1417,
+      "step": 3137
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 3e-05,
+      "loss": 0.1893,
+      "step": 3138
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 3e-05,
+      "loss": 0.1941,
+      "step": 3139
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 3e-05,
+      "loss": 0.1696,
+      "step": 3140
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 3e-05,
+      "loss": 0.1263,
+      "step": 3141
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 3e-05,
+      "loss": 0.3376,
+      "step": 3142
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 3e-05,
+      "loss": 0.16,
+      "step": 3143
+    },
+    {
+      "epoch": 25.77,
+      "learning_rate": 3e-05,
+      "loss": 0.1481,
+      "step": 3144
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 3e-05,
+      "loss": 0.1543,
+      "step": 3145
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 3e-05,
+      "loss": 0.2156,
+      "step": 3146
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1935,
+      "step": 3147
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1857,
+      "step": 3148
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 3e-05,
+      "loss": 0.1906,
+      "step": 3149
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 3e-05,
+      "loss": 0.201,
+      "step": 3150
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 3e-05,
+      "loss": 0.1685,
+      "step": 3151
+    },
+    {
+      "epoch": 25.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1609,
+      "step": 3152
+    },
+    {
+      "epoch": 25.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1122,
+      "step": 3153
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 3e-05,
+      "loss": 0.1686,
+      "step": 3154
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 3e-05,
+      "loss": 0.1339,
+      "step": 3155
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 3e-05,
+      "loss": 0.1432,
+      "step": 3156
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 3e-05,
+      "loss": 0.1713,
+      "step": 3157
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1249,
+      "step": 3158
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1613,
+      "step": 3159
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 3e-05,
+      "loss": 0.1697,
+      "step": 3160
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 3e-05,
+      "loss": 0.1882,
+      "step": 3161
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 3e-05,
+      "loss": 0.166,
+      "step": 3162
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 3e-05,
+      "loss": 0.153,
+      "step": 3163
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 3e-05,
+      "loss": 0.2365,
+      "step": 3164
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 3e-05,
+      "loss": 0.4219,
+      "step": 3165
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 3e-05,
+      "loss": 0.1956,
+      "step": 3166
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 3e-05,
+      "loss": 0.1732,
+      "step": 3167
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 3e-05,
+      "loss": 0.1791,
+      "step": 3168
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 3e-05,
+      "loss": 0.2019,
+      "step": 3169
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1239,
+      "step": 3170
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 3e-05,
+      "loss": 0.246,
+      "step": 3171
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 3e-05,
+      "loss": 0.0799,
+      "step": 3172
+    },
+    {
+      "epoch": 26.0,
+      "eval_loss": 0.273193359375,
+      "eval_mae": 0.38427734375,
+      "eval_mse": 0.273193359375,
+      "eval_runtime": 51.1417,
+      "eval_samples_per_second": 3234.018,
+      "eval_steps_per_second": 0.274,
+      "step": 3172
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 3e-05,
+      "loss": 0.1923,
+      "step": 3173
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 3e-05,
+      "loss": 0.2781,
+      "step": 3174
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1641,
+      "step": 3175
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 3e-05,
+      "loss": 0.1782,
+      "step": 3176
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 3e-05,
+      "loss": 0.24,
+      "step": 3177
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 3e-05,
+      "loss": 0.1873,
+      "step": 3178
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 3e-05,
+      "loss": 0.1776,
+      "step": 3179
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1544,
+      "step": 3180
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1267,
+      "step": 3181
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 3e-05,
+      "loss": 0.1433,
+      "step": 3182
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 3e-05,
+      "loss": 0.2037,
+      "step": 3183
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 3e-05,
+      "loss": 0.1759,
+      "step": 3184
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1793,
+      "step": 3185
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 3e-05,
+      "loss": 0.2129,
+      "step": 3186
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 3e-05,
+      "loss": 0.1309,
+      "step": 3187
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 3e-05,
+      "loss": 0.1753,
+      "step": 3188
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 3e-05,
+      "loss": 0.2349,
+      "step": 3189
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 3e-05,
+      "loss": 0.2063,
+      "step": 3190
+    },
+    {
+      "epoch": 26.16,
+      "learning_rate": 3e-05,
+      "loss": 0.146,
+      "step": 3191
+    },
+    {
+      "epoch": 26.16,
+      "learning_rate": 3e-05,
+      "loss": 0.2168,
+      "step": 3192
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 3e-05,
+      "loss": 0.3125,
+      "step": 3193
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 3e-05,
+      "loss": 0.1659,
+      "step": 3194
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 3e-05,
+      "loss": 0.2285,
+      "step": 3195
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1492,
+      "step": 3196
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 3e-05,
+      "loss": 0.4233,
+      "step": 3197
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 3e-05,
+      "loss": 0.1527,
+      "step": 3198
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 3e-05,
+      "loss": 0.1405,
+      "step": 3199
+    },
+    {
+      "epoch": 26.23,
+      "learning_rate": 3e-05,
+      "loss": 0.189,
+      "step": 3200
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 3e-05,
+      "loss": 0.1742,
+      "step": 3201
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 3e-05,
+      "loss": 0.2292,
+      "step": 3202
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1649,
+      "step": 3203
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 3e-05,
+      "loss": 0.1798,
+      "step": 3204
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 3e-05,
+      "loss": 0.1235,
+      "step": 3205
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 3e-05,
+      "loss": 0.1394,
+      "step": 3206
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 3e-05,
+      "loss": 0.3533,
+      "step": 3207
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 3e-05,
+      "loss": 0.2102,
+      "step": 3208
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1954,
+      "step": 3209
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 3e-05,
+      "loss": 0.1859,
+      "step": 3210
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 3e-05,
+      "loss": 0.2205,
+      "step": 3211
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 3e-05,
+      "loss": 0.3096,
+      "step": 3212
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 3e-05,
+      "loss": 0.2251,
+      "step": 3213
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 3e-05,
+      "loss": 0.2498,
+      "step": 3214
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 3e-05,
+      "loss": 0.1777,
+      "step": 3215
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 3e-05,
+      "loss": 0.2258,
+      "step": 3216
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 3e-05,
+      "loss": 0.2202,
+      "step": 3217
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 3e-05,
+      "loss": 0.1804,
+      "step": 3218
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1846,
+      "step": 3219
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 3e-05,
+      "loss": 0.2544,
+      "step": 3220
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 3e-05,
+      "loss": 0.114,
+      "step": 3221
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 3e-05,
+      "loss": 0.2336,
+      "step": 3222
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 3e-05,
+      "loss": 0.1462,
+      "step": 3223
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 3e-05,
+      "loss": 0.1763,
+      "step": 3224
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 3e-05,
+      "loss": 0.1393,
+      "step": 3225
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 3e-05,
+      "loss": 0.1841,
+      "step": 3226
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 3e-05,
+      "loss": 0.3284,
+      "step": 3227
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 3e-05,
+      "loss": 0.2177,
+      "step": 3228
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 3e-05,
+      "loss": 0.1473,
+      "step": 3229
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1901,
+      "step": 3230
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1583,
+      "step": 3231
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 3e-05,
+      "loss": 0.144,
+      "step": 3232
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 3e-05,
+      "loss": 0.2158,
+      "step": 3233
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 3e-05,
+      "loss": 0.1404,
+      "step": 3234
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1187,
+      "step": 3235
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 3e-05,
+      "loss": 0.2124,
+      "step": 3236
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 3e-05,
+      "loss": 0.1708,
+      "step": 3237
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 3e-05,
+      "loss": 0.2397,
+      "step": 3238
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 3e-05,
+      "loss": 0.1869,
+      "step": 3239
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 3e-05,
+      "loss": 0.1935,
+      "step": 3240
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 3e-05,
+      "loss": 0.2194,
+      "step": 3241
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 3e-05,
+      "loss": 0.2085,
+      "step": 3242
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 3e-05,
+      "loss": 0.157,
+      "step": 3243
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 3e-05,
+      "loss": 0.1488,
+      "step": 3244
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 3e-05,
+      "loss": 0.1521,
+      "step": 3245
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1841,
+      "step": 3246
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1105,
+      "step": 3247
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 3e-05,
+      "loss": 0.2605,
+      "step": 3248
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 3e-05,
+      "loss": 0.1653,
+      "step": 3249
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 3e-05,
+      "loss": 0.1869,
+      "step": 3250
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 3e-05,
+      "loss": 0.1633,
+      "step": 3251
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1981,
+      "step": 3252
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 3e-05,
+      "loss": 0.181,
+      "step": 3253
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 3e-05,
+      "loss": 0.2218,
+      "step": 3254
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 3e-05,
+      "loss": 0.1982,
+      "step": 3255
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 3e-05,
+      "loss": 0.1426,
+      "step": 3256
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1378,
+      "step": 3257
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 3e-05,
+      "loss": 0.142,
+      "step": 3258
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 3e-05,
+      "loss": 0.1268,
+      "step": 3259
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 3e-05,
+      "loss": 0.272,
+      "step": 3260
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 3e-05,
+      "loss": 0.2069,
+      "step": 3261
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 3e-05,
+      "loss": 0.2113,
+      "step": 3262
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 3e-05,
+      "loss": 0.2351,
+      "step": 3263
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 3e-05,
+      "loss": 0.196,
+      "step": 3264
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 3e-05,
+      "loss": 0.1943,
+      "step": 3265
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 3e-05,
+      "loss": 0.271,
+      "step": 3266
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 3e-05,
+      "loss": 0.1545,
+      "step": 3267
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 3e-05,
+      "loss": 0.3191,
+      "step": 3268
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1649,
+      "step": 3269
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 3e-05,
+      "loss": 0.3857,
+      "step": 3270
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 3e-05,
+      "loss": 0.1613,
+      "step": 3271
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 3e-05,
+      "loss": 0.2827,
+      "step": 3272
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 3e-05,
+      "loss": 0.1058,
+      "step": 3273
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 3e-05,
+      "loss": 0.2212,
+      "step": 3274
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 3e-05,
+      "loss": 0.2327,
+      "step": 3275
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 3e-05,
+      "loss": 0.174,
+      "step": 3276
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 3e-05,
+      "loss": 0.1725,
+      "step": 3277
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 3e-05,
+      "loss": 0.2507,
+      "step": 3278
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 3e-05,
+      "loss": 0.1523,
+      "step": 3279
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1669,
+      "step": 3280
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 3e-05,
+      "loss": 0.201,
+      "step": 3281
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 3e-05,
+      "loss": 0.171,
+      "step": 3282
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 3e-05,
+      "loss": 0.1604,
+      "step": 3283
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 3e-05,
+      "loss": 0.324,
+      "step": 3284
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 3e-05,
+      "loss": 0.16,
+      "step": 3285
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 3e-05,
+      "loss": 0.2517,
+      "step": 3286
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 3e-05,
+      "loss": 0.1428,
+      "step": 3287
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 3e-05,
+      "loss": 0.3335,
+      "step": 3288
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 3e-05,
+      "loss": 0.2032,
+      "step": 3289
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 3e-05,
+      "loss": 0.2593,
+      "step": 3290
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1101,
+      "step": 3291
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 3e-05,
+      "loss": 0.2478,
+      "step": 3292
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 3e-05,
+      "loss": 0.2563,
+      "step": 3293
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 3e-05,
+      "loss": 0.171,
+      "step": 3294
+    },
+    {
+      "epoch": 27.0,
+      "eval_loss": 0.271240234375,
+      "eval_mae": 0.3818359375,
+      "eval_mse": 0.271240234375,
+      "eval_runtime": 51.101,
+      "eval_samples_per_second": 3236.592,
+      "eval_steps_per_second": 0.274,
+      "step": 3294
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 3e-05,
+      "loss": 0.1736,
+      "step": 3295
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 3e-05,
+      "loss": 0.2096,
+      "step": 3296
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1477,
+      "step": 3297
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 3e-05,
+      "loss": 0.1774,
+      "step": 3298
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 3e-05,
+      "loss": 0.2013,
+      "step": 3299
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 3e-05,
+      "loss": 0.176,
+      "step": 3300
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 3e-05,
+      "loss": 0.1387,
+      "step": 3301
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 3e-05,
+      "loss": 0.123,
+      "step": 3302
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1497,
+      "step": 3303
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 3e-05,
+      "loss": 0.1774,
+      "step": 3304
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 3e-05,
+      "loss": 0.1682,
+      "step": 3305
+    },
+    {
+      "epoch": 27.1,
+      "learning_rate": 3e-05,
+      "loss": 0.1544,
+      "step": 3306
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1671,
+      "step": 3307
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1438,
+      "step": 3308
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 3e-05,
+      "loss": 0.1777,
+      "step": 3309
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 3e-05,
+      "loss": 0.2092,
+      "step": 3310
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 3e-05,
+      "loss": 0.2,
+      "step": 3311
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 3e-05,
+      "loss": 0.2404,
+      "step": 3312
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1104,
+      "step": 3313
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1155,
+      "step": 3314
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 3e-05,
+      "loss": 0.1569,
+      "step": 3315
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 3e-05,
+      "loss": 0.1186,
+      "step": 3316
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 3e-05,
+      "loss": 0.1589,
+      "step": 3317
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 3e-05,
+      "loss": 0.2693,
+      "step": 3318
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1893,
+      "step": 3319
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 3e-05,
+      "loss": 0.2465,
+      "step": 3320
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 3e-05,
+      "loss": 0.1447,
+      "step": 3321
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 3e-05,
+      "loss": 0.2178,
+      "step": 3322
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 3e-05,
+      "loss": 0.2072,
+      "step": 3323
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 3e-05,
+      "loss": 0.103,
+      "step": 3324
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1987,
+      "step": 3325
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 3e-05,
+      "loss": 0.155,
+      "step": 3326
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 3e-05,
+      "loss": 0.2041,
+      "step": 3327
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 3e-05,
+      "loss": 0.2305,
+      "step": 3328
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 3e-05,
+      "loss": 0.1757,
+      "step": 3329
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1744,
+      "step": 3330
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1296,
+      "step": 3331
+    },
+    {
+      "epoch": 27.31,
+      "learning_rate": 3e-05,
+      "loss": 0.1882,
+      "step": 3332
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 3e-05,
+      "loss": 0.1522,
+      "step": 3333
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 3e-05,
+      "loss": 0.2474,
+      "step": 3334
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 3e-05,
+      "loss": 0.3042,
+      "step": 3335
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 3e-05,
+      "loss": 0.2231,
+      "step": 3336
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 3e-05,
+      "loss": 0.1406,
+      "step": 3337
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 3e-05,
+      "loss": 0.1334,
+      "step": 3338
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 3e-05,
+      "loss": 0.1707,
+      "step": 3339
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 3e-05,
+      "loss": 0.1371,
+      "step": 3340
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1412,
+      "step": 3341
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1088,
+      "step": 3342
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 3e-05,
+      "loss": 0.1798,
+      "step": 3343
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 3e-05,
+      "loss": 0.3135,
+      "step": 3344
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 3e-05,
+      "loss": 0.1168,
+      "step": 3345
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 3e-05,
+      "loss": 0.2544,
+      "step": 3346
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 3e-05,
+      "loss": 0.2527,
+      "step": 3347
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 3e-05,
+      "loss": 0.1321,
+      "step": 3348
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 3e-05,
+      "loss": 0.2979,
+      "step": 3349
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 3e-05,
+      "loss": 0.2294,
+      "step": 3350
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 3e-05,
+      "loss": 0.2386,
+      "step": 3351
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1581,
+      "step": 3352
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1698,
+      "step": 3353
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 3e-05,
+      "loss": 0.1698,
+      "step": 3354
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 3e-05,
+      "loss": 0.2213,
+      "step": 3355
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 3e-05,
+      "loss": 0.1322,
+      "step": 3356
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1685,
+      "step": 3357
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1666,
+      "step": 3358
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 3e-05,
+      "loss": 0.2343,
+      "step": 3359
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 3e-05,
+      "loss": 0.1625,
+      "step": 3360
+    },
+    {
+      "epoch": 27.55,
+      "learning_rate": 3e-05,
+      "loss": 0.1427,
+      "step": 3361
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 3e-05,
+      "loss": 0.2059,
+      "step": 3362
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 3e-05,
+      "loss": 0.1449,
+      "step": 3363
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 3e-05,
+      "loss": 0.1049,
+      "step": 3364
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 3e-05,
+      "loss": 0.1888,
+      "step": 3365
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 3e-05,
+      "loss": 0.1588,
+      "step": 3366
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 3e-05,
+      "loss": 0.1523,
+      "step": 3367
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1787,
+      "step": 3368
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 3e-05,
+      "loss": 0.2345,
+      "step": 3369
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 3e-05,
+      "loss": 0.1473,
+      "step": 3370
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 3e-05,
+      "loss": 0.1399,
+      "step": 3371
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 3e-05,
+      "loss": 0.2212,
+      "step": 3372
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 3e-05,
+      "loss": 0.1925,
+      "step": 3373
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1855,
+      "step": 3374
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1409,
+      "step": 3375
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 3e-05,
+      "loss": 0.179,
+      "step": 3376
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 3e-05,
+      "loss": 0.1946,
+      "step": 3377
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 3e-05,
+      "loss": 0.1418,
+      "step": 3378
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 3e-05,
+      "loss": 0.4897,
+      "step": 3379
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1674,
+      "step": 3380
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 3e-05,
+      "loss": 0.1184,
+      "step": 3381
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 3e-05,
+      "loss": 0.1996,
+      "step": 3382
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 3e-05,
+      "loss": 0.1693,
+      "step": 3383
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 3e-05,
+      "loss": 0.2112,
+      "step": 3384
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 3e-05,
+      "loss": 0.202,
+      "step": 3385
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 3e-05,
+      "loss": 0.2211,
+      "step": 3386
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 3e-05,
+      "loss": 0.1748,
+      "step": 3387
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 3e-05,
+      "loss": 0.1549,
+      "step": 3388
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 3e-05,
+      "loss": 0.2358,
+      "step": 3389
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 3e-05,
+      "loss": 0.1444,
+      "step": 3390
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1273,
+      "step": 3391
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1456,
+      "step": 3392
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 3e-05,
+      "loss": 0.1403,
+      "step": 3393
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 3e-05,
+      "loss": 0.1687,
+      "step": 3394
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 3e-05,
+      "loss": 0.1426,
+      "step": 3395
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 3e-05,
+      "loss": 0.201,
+      "step": 3396
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1151,
+      "step": 3397
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 3e-05,
+      "loss": 0.1631,
+      "step": 3398
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 3e-05,
+      "loss": 0.1285,
+      "step": 3399
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 3e-05,
+      "loss": 0.1249,
+      "step": 3400
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 3e-05,
+      "loss": 0.152,
+      "step": 3401
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 3e-05,
+      "loss": 0.2198,
+      "step": 3402
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1449,
+      "step": 3403
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 3e-05,
+      "loss": 0.2556,
+      "step": 3404
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 3e-05,
+      "loss": 0.2316,
+      "step": 3405
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 3e-05,
+      "loss": 0.1377,
+      "step": 3406
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1758,
+      "step": 3407
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 3e-05,
+      "loss": 0.2554,
+      "step": 3408
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 3e-05,
+      "loss": 0.1808,
+      "step": 3409
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 3e-05,
+      "loss": 0.1422,
+      "step": 3410
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 3e-05,
+      "loss": 0.1942,
+      "step": 3411
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 3e-05,
+      "loss": 0.1865,
+      "step": 3412
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1753,
+      "step": 3413
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 3e-05,
+      "loss": 0.2002,
+      "step": 3414
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 3e-05,
+      "loss": 0.3191,
+      "step": 3415
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 3e-05,
+      "loss": 0.168,
+      "step": 3416
+    },
+    {
+      "epoch": 28.0,
+      "eval_loss": 0.26953125,
+      "eval_mae": 0.382080078125,
+      "eval_mse": 0.26953125,
+      "eval_runtime": 51.1152,
+      "eval_samples_per_second": 3235.689,
+      "eval_steps_per_second": 0.274,
+      "step": 3416
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 3e-05,
+      "loss": 0.1368,
+      "step": 3417
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1301,
+      "step": 3418
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 3e-05,
+      "loss": 0.2108,
+      "step": 3419
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 3e-05,
+      "loss": 0.1492,
+      "step": 3420
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 3e-05,
+      "loss": 0.1212,
+      "step": 3421
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 3e-05,
+      "loss": 0.1406,
+      "step": 3422
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 3e-05,
+      "loss": 0.1376,
+      "step": 3423
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 3e-05,
+      "loss": 0.2045,
+      "step": 3424
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 3e-05,
+      "loss": 0.2947,
+      "step": 3425
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 3e-05,
+      "loss": 0.2128,
+      "step": 3426
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 3e-05,
+      "loss": 0.1216,
+      "step": 3427
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 3e-05,
+      "loss": 0.1877,
+      "step": 3428
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1338,
+      "step": 3429
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1177,
+      "step": 3430
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 3e-05,
+      "loss": 0.2074,
+      "step": 3431
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 3e-05,
+      "loss": 0.1472,
+      "step": 3432
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 3e-05,
+      "loss": 0.1555,
+      "step": 3433
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 3e-05,
+      "loss": 0.1578,
+      "step": 3434
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 3e-05,
+      "loss": 0.2305,
+      "step": 3435
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1401,
+      "step": 3436
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 3e-05,
+      "loss": 0.1421,
+      "step": 3437
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 3e-05,
+      "loss": 0.2032,
+      "step": 3438
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 3e-05,
+      "loss": 0.1852,
+      "step": 3439
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1949,
+      "step": 3440
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1882,
+      "step": 3441
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 3e-05,
+      "loss": 0.1927,
+      "step": 3442
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 3e-05,
+      "loss": 0.2305,
+      "step": 3443
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 3e-05,
+      "loss": 0.1871,
+      "step": 3444
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 3e-05,
+      "loss": 0.1973,
+      "step": 3445
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1813,
+      "step": 3446
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1065,
+      "step": 3447
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 3e-05,
+      "loss": 0.1512,
+      "step": 3448
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 3e-05,
+      "loss": 0.3257,
+      "step": 3449
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 3e-05,
+      "loss": 0.1318,
+      "step": 3450
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 3e-05,
+      "loss": 0.2585,
+      "step": 3451
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1691,
+      "step": 3452
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1296,
+      "step": 3453
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 3e-05,
+      "loss": 0.1484,
+      "step": 3454
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 3e-05,
+      "loss": 0.1942,
+      "step": 3455
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 3e-05,
+      "loss": 0.1744,
+      "step": 3456
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1399,
+      "step": 3457
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1733,
+      "step": 3458
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 3e-05,
+      "loss": 0.1628,
+      "step": 3459
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 3e-05,
+      "loss": 0.2006,
+      "step": 3460
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 3e-05,
+      "loss": 0.1398,
+      "step": 3461
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 3e-05,
+      "loss": 0.1175,
+      "step": 3462
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1567,
+      "step": 3463
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1865,
+      "step": 3464
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 3e-05,
+      "loss": 0.2131,
+      "step": 3465
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 3e-05,
+      "loss": 0.1572,
+      "step": 3466
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 3e-05,
+      "loss": 0.1342,
+      "step": 3467
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 3e-05,
+      "loss": 0.1804,
+      "step": 3468
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 3e-05,
+      "loss": 0.225,
+      "step": 3469
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 3e-05,
+      "loss": 0.1254,
+      "step": 3470
+    },
+    {
+      "epoch": 28.45,
+      "learning_rate": 3e-05,
+      "loss": 0.1685,
+      "step": 3471
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 3e-05,
+      "loss": 0.1768,
+      "step": 3472
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 3e-05,
+      "loss": 0.2578,
+      "step": 3473
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1945,
+      "step": 3474
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 3e-05,
+      "loss": 0.2424,
+      "step": 3475
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 3e-05,
+      "loss": 0.1481,
+      "step": 3476
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 3e-05,
+      "loss": 0.1405,
+      "step": 3477
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 3e-05,
+      "loss": 0.1957,
+      "step": 3478
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 3e-05,
+      "loss": 0.2859,
+      "step": 3479
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1943,
+      "step": 3480
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 3e-05,
+      "loss": 0.2228,
+      "step": 3481
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 3e-05,
+      "loss": 0.248,
+      "step": 3482
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 3e-05,
+      "loss": 0.1338,
+      "step": 3483
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 3e-05,
+      "loss": 0.1611,
+      "step": 3484
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 3e-05,
+      "loss": 0.2196,
+      "step": 3485
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 3e-05,
+      "loss": 0.1458,
+      "step": 3486
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 3e-05,
+      "loss": 0.2064,
+      "step": 3487
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 3e-05,
+      "loss": 0.1746,
+      "step": 3488
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 3e-05,
+      "loss": 0.1487,
+      "step": 3489
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1522,
+      "step": 3490
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1343,
+      "step": 3491
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 3e-05,
+      "loss": 0.1525,
+      "step": 3492
+    },
+    {
+      "epoch": 28.63,
+      "learning_rate": 3e-05,
+      "loss": 0.1266,
+      "step": 3493
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 3e-05,
+      "loss": 0.1578,
+      "step": 3494
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 3e-05,
+      "loss": 0.1504,
+      "step": 3495
+    },
+    {
+      "epoch": 28.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1106,
+      "step": 3496
+    },
+    {
+      "epoch": 28.66,
+      "learning_rate": 3e-05,
+      "loss": 0.0837,
+      "step": 3497
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 3e-05,
+      "loss": 0.2006,
+      "step": 3498
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 3e-05,
+      "loss": 0.1902,
+      "step": 3499
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 3e-05,
+      "loss": 0.2715,
+      "step": 3500
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1534,
+      "step": 3501
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 3e-05,
+      "loss": 0.233,
+      "step": 3502
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 3e-05,
+      "loss": 0.2212,
+      "step": 3503
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 3e-05,
+      "loss": 0.1818,
+      "step": 3504
+    },
+    {
+      "epoch": 28.73,
+      "learning_rate": 3e-05,
+      "loss": 0.14,
+      "step": 3505
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 3e-05,
+      "loss": 0.1772,
+      "step": 3506
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 3e-05,
+      "loss": 0.2529,
+      "step": 3507
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 3e-05,
+      "loss": 0.172,
+      "step": 3508
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 3e-05,
+      "loss": 0.1527,
+      "step": 3509
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 3e-05,
+      "loss": 0.1769,
+      "step": 3510
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 3e-05,
+      "loss": 0.1881,
+      "step": 3511
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 3e-05,
+      "loss": 0.1624,
+      "step": 3512
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 3e-05,
+      "loss": 0.129,
+      "step": 3513
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 3e-05,
+      "loss": 0.2053,
+      "step": 3514
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 3e-05,
+      "loss": 0.1675,
+      "step": 3515
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 3e-05,
+      "loss": 0.2123,
+      "step": 3516
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 3e-05,
+      "loss": 0.1985,
+      "step": 3517
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1788,
+      "step": 3518
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 3e-05,
+      "loss": 0.2515,
+      "step": 3519
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 3e-05,
+      "loss": 0.1986,
+      "step": 3520
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 3e-05,
+      "loss": 0.1985,
+      "step": 3521
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 3e-05,
+      "loss": 0.2207,
+      "step": 3522
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 3e-05,
+      "loss": 0.1959,
+      "step": 3523
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 3e-05,
+      "loss": 0.3096,
+      "step": 3524
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1487,
+      "step": 3525
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 3e-05,
+      "loss": 0.1465,
+      "step": 3526
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 3e-05,
+      "loss": 0.1794,
+      "step": 3527
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 3e-05,
+      "loss": 0.1675,
+      "step": 3528
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 3e-05,
+      "loss": 0.2313,
+      "step": 3529
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 3e-05,
+      "loss": 0.224,
+      "step": 3530
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 3e-05,
+      "loss": 0.1818,
+      "step": 3531
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 3e-05,
+      "loss": 0.149,
+      "step": 3532
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 3e-05,
+      "loss": 0.1898,
+      "step": 3533
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 3e-05,
+      "loss": 0.1765,
+      "step": 3534
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1124,
+      "step": 3535
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1627,
+      "step": 3536
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 3e-05,
+      "loss": 0.1603,
+      "step": 3537
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 3e-05,
+      "loss": 0.1997,
+      "step": 3538
+    },
+    {
+      "epoch": 29.0,
+      "eval_loss": 0.2685546875,
+      "eval_mae": 0.380126953125,
+      "eval_mse": 0.2685546875,
+      "eval_runtime": 51.0746,
+      "eval_samples_per_second": 3238.262,
+      "eval_steps_per_second": 0.274,
+      "step": 3538
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 3e-05,
+      "loss": 0.1521,
+      "step": 3539
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1753,
+      "step": 3540
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1995,
+      "step": 3541
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 3e-05,
+      "loss": 0.1561,
+      "step": 3542
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 3e-05,
+      "loss": 0.155,
+      "step": 3543
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 3e-05,
+      "loss": 0.2256,
+      "step": 3544
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 3e-05,
+      "loss": 0.259,
+      "step": 3545
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 3e-05,
+      "loss": 0.2111,
+      "step": 3546
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1777,
+      "step": 3547
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 3e-05,
+      "loss": 0.2264,
+      "step": 3548
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 3e-05,
+      "loss": 0.2563,
+      "step": 3549
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 3e-05,
+      "loss": 0.2837,
+      "step": 3550
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 3e-05,
+      "loss": 0.177,
+      "step": 3551
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1536,
+      "step": 3552
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 3e-05,
+      "loss": 0.1523,
+      "step": 3553
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 3e-05,
+      "loss": 0.1415,
+      "step": 3554
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 3e-05,
+      "loss": 0.1506,
+      "step": 3555
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 3e-05,
+      "loss": 0.2074,
+      "step": 3556
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1965,
+      "step": 3557
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1399,
+      "step": 3558
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 3e-05,
+      "loss": 0.1799,
+      "step": 3559
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 3e-05,
+      "loss": 0.1855,
+      "step": 3560
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 3e-05,
+      "loss": 0.1592,
+      "step": 3561
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 3e-05,
+      "loss": 0.2886,
+      "step": 3562
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 3e-05,
+      "loss": 0.2952,
+      "step": 3563
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 3e-05,
+      "loss": 0.1459,
+      "step": 3564
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 3e-05,
+      "loss": 0.1797,
+      "step": 3565
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 3e-05,
+      "loss": 0.2283,
+      "step": 3566
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 3e-05,
+      "loss": 0.1025,
+      "step": 3567
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1135,
+      "step": 3568
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1676,
+      "step": 3569
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 3e-05,
+      "loss": 0.1235,
+      "step": 3570
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 3e-05,
+      "loss": 0.1769,
+      "step": 3571
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 3e-05,
+      "loss": 0.1209,
+      "step": 3572
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 3e-05,
+      "loss": 0.168,
+      "step": 3573
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 3e-05,
+      "loss": 0.212,
+      "step": 3574
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1158,
+      "step": 3575
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 3e-05,
+      "loss": 0.1476,
+      "step": 3576
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 3e-05,
+      "loss": 0.1437,
+      "step": 3577
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 3e-05,
+      "loss": 0.1082,
+      "step": 3578
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 3e-05,
+      "loss": 0.2842,
+      "step": 3579
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1345,
+      "step": 3580
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 3e-05,
+      "loss": 0.1423,
+      "step": 3581
+    },
+    {
+      "epoch": 29.36,
+      "learning_rate": 3e-05,
+      "loss": 0.2168,
+      "step": 3582
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 3e-05,
+      "loss": 0.1814,
+      "step": 3583
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 3e-05,
+      "loss": 0.1565,
+      "step": 3584
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1842,
+      "step": 3585
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 3e-05,
+      "loss": 0.2257,
+      "step": 3586
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 3e-05,
+      "loss": 0.1809,
+      "step": 3587
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 3e-05,
+      "loss": 0.2041,
+      "step": 3588
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 3e-05,
+      "loss": 0.178,
+      "step": 3589
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 3e-05,
+      "loss": 0.2159,
+      "step": 3590
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 3e-05,
+      "loss": 0.231,
+      "step": 3591
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 3e-05,
+      "loss": 0.2057,
+      "step": 3592
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 3e-05,
+      "loss": 0.2026,
+      "step": 3593
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 3e-05,
+      "loss": 0.1197,
+      "step": 3594
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 3e-05,
+      "loss": 0.1829,
+      "step": 3595
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 3e-05,
+      "loss": 0.2455,
+      "step": 3596
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 3e-05,
+      "loss": 0.2184,
+      "step": 3597
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 3e-05,
+      "loss": 0.1431,
+      "step": 3598
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 3e-05,
+      "loss": 0.1047,
+      "step": 3599
+    },
+    {
+      "epoch": 29.51,
+      "learning_rate": 3e-05,
+      "loss": 0.1798,
+      "step": 3600
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 3e-05,
+      "loss": 0.259,
+      "step": 3601
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1847,
+      "step": 3602
+    },
+    {
+      "epoch": 29.53,
+      "learning_rate": 3e-05,
+      "loss": 0.1376,
+      "step": 3603
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 3e-05,
+      "loss": 0.2402,
+      "step": 3604
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 3e-05,
+      "loss": 0.2029,
+      "step": 3605
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 3e-05,
+      "loss": 0.1025,
+      "step": 3606
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 3e-05,
+      "loss": 0.1836,
+      "step": 3607
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 3e-05,
+      "loss": 0.1852,
+      "step": 3608
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 3e-05,
+      "loss": 0.1864,
+      "step": 3609
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 3e-05,
+      "loss": 0.1122,
+      "step": 3610
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 3e-05,
+      "loss": 0.1586,
+      "step": 3611
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1765,
+      "step": 3612
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1378,
+      "step": 3613
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 3e-05,
+      "loss": 0.1885,
+      "step": 3614
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 3e-05,
+      "loss": 0.1715,
+      "step": 3615
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 3e-05,
+      "loss": 0.1124,
+      "step": 3616
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 3e-05,
+      "loss": 0.1687,
+      "step": 3617
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 3e-05,
+      "loss": 0.2888,
+      "step": 3618
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1827,
+      "step": 3619
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 3e-05,
+      "loss": 0.1846,
+      "step": 3620
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 3e-05,
+      "loss": 0.1309,
+      "step": 3621
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 3e-05,
+      "loss": 0.1576,
+      "step": 3622
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1278,
+      "step": 3623
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1214,
+      "step": 3624
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 3e-05,
+      "loss": 0.2075,
+      "step": 3625
+    },
+    {
+      "epoch": 29.72,
+      "learning_rate": 3e-05,
+      "loss": 0.1932,
+      "step": 3626
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 3e-05,
+      "loss": 0.1902,
+      "step": 3627
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 3e-05,
+      "loss": 0.1519,
+      "step": 3628
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 3e-05,
+      "loss": 0.1625,
+      "step": 3629
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 3e-05,
+      "loss": 0.1478,
+      "step": 3630
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 3e-05,
+      "loss": 0.165,
+      "step": 3631
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 3e-05,
+      "loss": 0.1987,
+      "step": 3632
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 3e-05,
+      "loss": 0.1324,
+      "step": 3633
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 3e-05,
+      "loss": 0.1133,
+      "step": 3634
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1287,
+      "step": 3635
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1633,
+      "step": 3636
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 3e-05,
+      "loss": 0.2386,
+      "step": 3637
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 3e-05,
+      "loss": 0.1335,
+      "step": 3638
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 3e-05,
+      "loss": 0.2136,
+      "step": 3639
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1924,
+      "step": 3640
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1709,
+      "step": 3641
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 3e-05,
+      "loss": 0.1414,
+      "step": 3642
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 3e-05,
+      "loss": 0.2766,
+      "step": 3643
+    },
+    {
+      "epoch": 29.87,
+      "learning_rate": 3e-05,
+      "loss": 0.1519,
+      "step": 3644
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 3e-05,
+      "loss": 0.17,
+      "step": 3645
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1892,
+      "step": 3646
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1996,
+      "step": 3647
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 3e-05,
+      "loss": 0.222,
+      "step": 3648
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 3e-05,
+      "loss": 0.1136,
+      "step": 3649
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 3e-05,
+      "loss": 0.199,
+      "step": 3650
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1066,
+      "step": 3651
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1434,
+      "step": 3652
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 3e-05,
+      "loss": 0.1534,
+      "step": 3653
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 3e-05,
+      "loss": 0.1968,
+      "step": 3654
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 3e-05,
+      "loss": 0.1322,
+      "step": 3655
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 3e-05,
+      "loss": 0.1744,
+      "step": 3656
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1963,
+      "step": 3657
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1692,
+      "step": 3658
+    },
+    {
+      "epoch": 29.99,
+      "learning_rate": 3e-05,
+      "loss": 0.142,
+      "step": 3659
+    },
+    {
+      "epoch": 30.0,
+      "learning_rate": 3e-05,
+      "loss": 0.3416,
+      "step": 3660
+    },
+    {
+      "epoch": 30.0,
+      "eval_loss": 0.263671875,
+      "eval_mae": 0.37451171875,
+      "eval_mse": 0.263671875,
+      "eval_runtime": 51.1349,
+      "eval_samples_per_second": 3234.442,
+      "eval_steps_per_second": 0.274,
+      "step": 3660
+    },
+    {
+      "epoch": 30.01,
+      "learning_rate": 3e-05,
+      "loss": 0.1222,
+      "step": 3661
+    },
+    {
+      "epoch": 30.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1536,
+      "step": 3662
+    },
+    {
+      "epoch": 30.02,
+      "learning_rate": 3e-05,
+      "loss": 0.226,
+      "step": 3663
+    },
+    {
+      "epoch": 30.03,
+      "learning_rate": 3e-05,
+      "loss": 0.1688,
+      "step": 3664
+    },
+    {
+      "epoch": 30.04,
+      "learning_rate": 3e-05,
+      "loss": 0.1471,
+      "step": 3665
+    },
+    {
+      "epoch": 30.05,
+      "learning_rate": 3e-05,
+      "loss": 0.1238,
+      "step": 3666
+    },
+    {
+      "epoch": 30.06,
+      "learning_rate": 3e-05,
+      "loss": 0.1373,
+      "step": 3667
+    },
+    {
+      "epoch": 30.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1938,
+      "step": 3668
+    },
+    {
+      "epoch": 30.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1587,
+      "step": 3669
+    },
+    {
+      "epoch": 30.08,
+      "learning_rate": 3e-05,
+      "loss": 0.0934,
+      "step": 3670
+    },
+    {
+      "epoch": 30.09,
+      "learning_rate": 3e-05,
+      "loss": 0.1743,
+      "step": 3671
+    },
+    {
+      "epoch": 30.1,
+      "learning_rate": 3e-05,
+      "loss": 0.1552,
+      "step": 3672
+    },
+    {
+      "epoch": 30.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1979,
+      "step": 3673
+    },
+    {
+      "epoch": 30.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1492,
+      "step": 3674
+    },
+    {
+      "epoch": 30.12,
+      "learning_rate": 3e-05,
+      "loss": 0.184,
+      "step": 3675
+    },
+    {
+      "epoch": 30.13,
+      "learning_rate": 3e-05,
+      "loss": 0.2333,
+      "step": 3676
+    },
+    {
+      "epoch": 30.14,
+      "learning_rate": 3e-05,
+      "loss": 0.1205,
+      "step": 3677
+    },
+    {
+      "epoch": 30.15,
+      "learning_rate": 3e-05,
+      "loss": 0.2241,
+      "step": 3678
+    },
+    {
+      "epoch": 30.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1366,
+      "step": 3679
+    },
+    {
+      "epoch": 30.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1433,
+      "step": 3680
+    },
+    {
+      "epoch": 30.17,
+      "learning_rate": 3e-05,
+      "loss": 0.2402,
+      "step": 3681
+    },
+    {
+      "epoch": 30.18,
+      "learning_rate": 3e-05,
+      "loss": 0.1746,
+      "step": 3682
+    },
+    {
+      "epoch": 30.19,
+      "learning_rate": 3e-05,
+      "loss": 0.1556,
+      "step": 3683
+    },
+    {
+      "epoch": 30.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1801,
+      "step": 3684
+    },
+    {
+      "epoch": 30.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1459,
+      "step": 3685
+    },
+    {
+      "epoch": 30.21,
+      "learning_rate": 3e-05,
+      "loss": 0.2068,
+      "step": 3686
+    },
+    {
+      "epoch": 30.22,
+      "learning_rate": 3e-05,
+      "loss": 0.2881,
+      "step": 3687
+    },
+    {
+      "epoch": 30.23,
+      "learning_rate": 3e-05,
+      "loss": 0.1388,
+      "step": 3688
+    },
+    {
+      "epoch": 30.24,
+      "learning_rate": 3e-05,
+      "loss": 0.1577,
+      "step": 3689
+    },
+    {
+      "epoch": 30.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1421,
+      "step": 3690
+    },
+    {
+      "epoch": 30.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1547,
+      "step": 3691
+    },
+    {
+      "epoch": 30.26,
+      "learning_rate": 3e-05,
+      "loss": 0.2191,
+      "step": 3692
+    },
+    {
+      "epoch": 30.27,
+      "learning_rate": 3e-05,
+      "loss": 0.1675,
+      "step": 3693
+    },
+    {
+      "epoch": 30.28,
+      "learning_rate": 3e-05,
+      "loss": 0.109,
+      "step": 3694
+    },
+    {
+      "epoch": 30.29,
+      "learning_rate": 3e-05,
+      "loss": 0.1674,
+      "step": 3695
+    },
+    {
+      "epoch": 30.3,
+      "learning_rate": 3e-05,
+      "loss": 0.192,
+      "step": 3696
+    },
+    {
+      "epoch": 30.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1458,
+      "step": 3697
+    },
+    {
+      "epoch": 30.31,
+      "learning_rate": 3e-05,
+      "loss": 0.1608,
+      "step": 3698
+    },
+    {
+      "epoch": 30.32,
+      "learning_rate": 3e-05,
+      "loss": 0.1469,
+      "step": 3699
+    },
+    {
+      "epoch": 30.33,
+      "learning_rate": 3e-05,
+      "loss": 0.1102,
+      "step": 3700
+    },
+    {
+      "epoch": 30.34,
+      "learning_rate": 3e-05,
+      "loss": 0.2316,
+      "step": 3701
+    },
+    {
+      "epoch": 30.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1683,
+      "step": 3702
+    },
+    {
+      "epoch": 30.35,
+      "learning_rate": 3e-05,
+      "loss": 0.1552,
+      "step": 3703
+    },
+    {
+      "epoch": 30.36,
+      "learning_rate": 3e-05,
+      "loss": 0.17,
+      "step": 3704
+    },
+    {
+      "epoch": 30.37,
+      "learning_rate": 3e-05,
+      "loss": 0.1537,
+      "step": 3705
+    },
+    {
+      "epoch": 30.38,
+      "learning_rate": 3e-05,
+      "loss": 0.1543,
+      "step": 3706
+    },
+    {
+      "epoch": 30.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1122,
+      "step": 3707
+    },
+    {
+      "epoch": 30.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1637,
+      "step": 3708
+    },
+    {
+      "epoch": 30.4,
+      "learning_rate": 3e-05,
+      "loss": 0.2164,
+      "step": 3709
+    },
+    {
+      "epoch": 30.41,
+      "learning_rate": 3e-05,
+      "loss": 0.191,
+      "step": 3710
+    },
+    {
+      "epoch": 30.42,
+      "learning_rate": 3e-05,
+      "loss": 0.1,
+      "step": 3711
+    },
+    {
+      "epoch": 30.43,
+      "learning_rate": 3e-05,
+      "loss": 0.2717,
+      "step": 3712
+    },
+    {
+      "epoch": 30.43,
+      "learning_rate": 3e-05,
+      "loss": 0.2952,
+      "step": 3713
+    },
+    {
+      "epoch": 30.44,
+      "learning_rate": 3e-05,
+      "loss": 0.2793,
+      "step": 3714
+    },
+    {
+      "epoch": 30.45,
+      "learning_rate": 3e-05,
+      "loss": 0.152,
+      "step": 3715
+    },
+    {
+      "epoch": 30.46,
+      "learning_rate": 3e-05,
+      "loss": 0.1097,
+      "step": 3716
+    },
+    {
+      "epoch": 30.47,
+      "learning_rate": 3e-05,
+      "loss": 0.1757,
+      "step": 3717
+    },
+    {
+      "epoch": 30.48,
+      "learning_rate": 3e-05,
+      "loss": 0.2461,
+      "step": 3718
+    },
+    {
+      "epoch": 30.48,
+      "learning_rate": 3e-05,
+      "loss": 0.228,
+      "step": 3719
+    },
+    {
+      "epoch": 30.49,
+      "learning_rate": 3e-05,
+      "loss": 0.1588,
+      "step": 3720
+    },
+    {
+      "epoch": 30.5,
+      "learning_rate": 3e-05,
+      "loss": 0.2073,
+      "step": 3721
+    },
+    {
+      "epoch": 30.51,
+      "learning_rate": 3e-05,
+      "loss": 0.1504,
+      "step": 3722
+    },
+    {
+      "epoch": 30.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1476,
+      "step": 3723
+    },
+    {
+      "epoch": 30.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1517,
+      "step": 3724
+    },
+    {
+      "epoch": 30.53,
+      "learning_rate": 3e-05,
+      "loss": 0.2244,
+      "step": 3725
+    },
+    {
+      "epoch": 30.54,
+      "learning_rate": 3e-05,
+      "loss": 0.1678,
+      "step": 3726
+    },
+    {
+      "epoch": 30.55,
+      "learning_rate": 3e-05,
+      "loss": 0.1534,
+      "step": 3727
+    },
+    {
+      "epoch": 30.56,
+      "learning_rate": 3e-05,
+      "loss": 0.1643,
+      "step": 3728
+    },
+    {
+      "epoch": 30.57,
+      "learning_rate": 3e-05,
+      "loss": 0.1107,
+      "step": 3729
+    },
+    {
+      "epoch": 30.57,
+      "learning_rate": 3e-05,
+      "loss": 0.2056,
+      "step": 3730
+    },
+    {
+      "epoch": 30.58,
+      "learning_rate": 3e-05,
+      "loss": 0.2441,
+      "step": 3731
+    },
+    {
+      "epoch": 30.59,
+      "learning_rate": 3e-05,
+      "loss": 0.124,
+      "step": 3732
+    },
+    {
+      "epoch": 30.6,
+      "learning_rate": 3e-05,
+      "loss": 0.1539,
+      "step": 3733
+    },
+    {
+      "epoch": 30.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1958,
+      "step": 3734
+    },
+    {
+      "epoch": 30.61,
+      "learning_rate": 3e-05,
+      "loss": 0.3535,
+      "step": 3735
+    },
+    {
+      "epoch": 30.62,
+      "learning_rate": 3e-05,
+      "loss": 0.119,
+      "step": 3736
+    },
+    {
+      "epoch": 30.63,
+      "learning_rate": 3e-05,
+      "loss": 0.1947,
+      "step": 3737
+    },
+    {
+      "epoch": 30.64,
+      "learning_rate": 3e-05,
+      "loss": 0.1357,
+      "step": 3738
+    },
+    {
+      "epoch": 30.65,
+      "learning_rate": 3e-05,
+      "loss": 0.14,
+      "step": 3739
+    },
+    {
+      "epoch": 30.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1255,
+      "step": 3740
+    },
+    {
+      "epoch": 30.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1294,
+      "step": 3741
+    },
+    {
+      "epoch": 30.67,
+      "learning_rate": 3e-05,
+      "loss": 0.1947,
+      "step": 3742
+    },
+    {
+      "epoch": 30.68,
+      "learning_rate": 3e-05,
+      "loss": 0.1598,
+      "step": 3743
+    },
+    {
+      "epoch": 30.69,
+      "learning_rate": 3e-05,
+      "loss": 0.173,
+      "step": 3744
+    },
+    {
+      "epoch": 30.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1978,
+      "step": 3745
+    },
+    {
+      "epoch": 30.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1154,
+      "step": 3746
+    },
+    {
+      "epoch": 30.71,
+      "learning_rate": 3e-05,
+      "loss": 0.1599,
+      "step": 3747
+    },
+    {
+      "epoch": 30.72,
+      "learning_rate": 3e-05,
+      "loss": 0.1707,
+      "step": 3748
+    },
+    {
+      "epoch": 30.73,
+      "learning_rate": 3e-05,
+      "loss": 0.184,
+      "step": 3749
+    },
+    {
+      "epoch": 30.74,
+      "learning_rate": 3e-05,
+      "loss": 0.161,
+      "step": 3750
+    },
+    {
+      "epoch": 30.75,
+      "learning_rate": 3e-05,
+      "loss": 0.1562,
+      "step": 3751
+    },
+    {
+      "epoch": 30.75,
+      "learning_rate": 3e-05,
+      "loss": 0.1677,
+      "step": 3752
+    },
+    {
+      "epoch": 30.76,
+      "learning_rate": 3e-05,
+      "loss": 0.1743,
+      "step": 3753
+    },
+    {
+      "epoch": 30.77,
+      "learning_rate": 3e-05,
+      "loss": 0.1488,
+      "step": 3754
+    },
+    {
+      "epoch": 30.78,
+      "learning_rate": 3e-05,
+      "loss": 0.3369,
+      "step": 3755
+    },
+    {
+      "epoch": 30.79,
+      "learning_rate": 3e-05,
+      "loss": 0.1945,
+      "step": 3756
+    },
+    {
+      "epoch": 30.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1101,
+      "step": 3757
+    },
+    {
+      "epoch": 30.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1115,
+      "step": 3758
+    },
+    {
+      "epoch": 30.81,
+      "learning_rate": 3e-05,
+      "loss": 0.2399,
+      "step": 3759
+    },
+    {
+      "epoch": 30.82,
+      "learning_rate": 3e-05,
+      "loss": 0.1558,
+      "step": 3760
+    },
+    {
+      "epoch": 30.83,
+      "learning_rate": 3e-05,
+      "loss": 0.2019,
+      "step": 3761
+    },
+    {
+      "epoch": 30.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1564,
+      "step": 3762
+    },
+    {
+      "epoch": 30.84,
+      "learning_rate": 3e-05,
+      "loss": 0.2018,
+      "step": 3763
+    },
+    {
+      "epoch": 30.85,
+      "learning_rate": 3e-05,
+      "loss": 0.2205,
+      "step": 3764
+    },
+    {
+      "epoch": 30.86,
+      "learning_rate": 3e-05,
+      "loss": 0.1768,
+      "step": 3765
+    },
+    {
+      "epoch": 30.87,
+      "learning_rate": 3e-05,
+      "loss": 0.2104,
+      "step": 3766
+    },
+    {
+      "epoch": 30.88,
+      "learning_rate": 3e-05,
+      "loss": 0.313,
+      "step": 3767
+    },
+    {
+      "epoch": 30.89,
+      "learning_rate": 3e-05,
+      "loss": 0.107,
+      "step": 3768
+    },
+    {
+      "epoch": 30.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1403,
+      "step": 3769
+    },
+    {
+      "epoch": 30.9,
+      "learning_rate": 3e-05,
+      "loss": 0.1307,
+      "step": 3770
+    },
+    {
+      "epoch": 30.91,
+      "learning_rate": 3e-05,
+      "loss": 0.0866,
+      "step": 3771
+    },
+    {
+      "epoch": 30.92,
+      "learning_rate": 3e-05,
+      "loss": 0.3965,
+      "step": 3772
+    },
+    {
+      "epoch": 30.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1066,
+      "step": 3773
+    },
+    {
+      "epoch": 30.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1306,
+      "step": 3774
+    },
+    {
+      "epoch": 30.94,
+      "learning_rate": 3e-05,
+      "loss": 0.1481,
+      "step": 3775
+    },
+    {
+      "epoch": 30.95,
+      "learning_rate": 3e-05,
+      "loss": 0.1921,
+      "step": 3776
+    },
+    {
+      "epoch": 30.96,
+      "learning_rate": 3e-05,
+      "loss": 0.13,
+      "step": 3777
+    },
+    {
+      "epoch": 30.97,
+      "learning_rate": 3e-05,
+      "loss": 0.1631,
+      "step": 3778
+    },
+    {
+      "epoch": 30.98,
+      "learning_rate": 3e-05,
+      "loss": 0.239,
+      "step": 3779
+    },
+    {
+      "epoch": 30.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1164,
+      "step": 3780
+    },
+    {
+      "epoch": 30.99,
+      "learning_rate": 3e-05,
+      "loss": 0.1405,
+      "step": 3781
+    },
+    {
+      "epoch": 31.0,
+      "learning_rate": 3e-05,
+      "loss": 0.0898,
+      "step": 3782
+    },
+    {
+      "epoch": 31.0,
+      "eval_loss": 0.265869140625,
+      "eval_mae": 0.375732421875,
+      "eval_mse": 0.265869140625,
+      "eval_runtime": 51.1152,
+      "eval_samples_per_second": 3235.692,
+      "eval_steps_per_second": 0.274,
+      "step": 3782
+    },
+    {
+      "epoch": 31.01,
+      "learning_rate": 3e-05,
+      "loss": 0.1793,
+      "step": 3783
+    },
+    {
+      "epoch": 31.02,
+      "learning_rate": 3e-05,
+      "loss": 0.158,
+      "step": 3784
+    },
+    {
+      "epoch": 31.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1886,
+      "step": 3785
+    },
+    {
+      "epoch": 31.03,
+      "learning_rate": 3e-05,
+      "loss": 0.2406,
+      "step": 3786
+    },
+    {
+      "epoch": 31.04,
+      "learning_rate": 3e-05,
+      "loss": 0.1398,
+      "step": 3787
+    },
+    {
+      "epoch": 31.05,
+      "learning_rate": 3e-05,
+      "loss": 0.16,
+      "step": 3788
+    },
+    {
+      "epoch": 31.06,
+      "learning_rate": 3e-05,
+      "loss": 0.1877,
+      "step": 3789
+    },
+    {
+      "epoch": 31.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1318,
+      "step": 3790
+    },
+    {
+      "epoch": 31.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1545,
+      "step": 3791
+    },
+    {
+      "epoch": 31.08,
+      "learning_rate": 3e-05,
+      "loss": 0.1785,
+      "step": 3792
+    },
+    {
+      "epoch": 31.09,
+      "learning_rate": 3e-05,
+      "loss": 0.2404,
+      "step": 3793
+    },
+    {
+      "epoch": 31.1,
+      "learning_rate": 3e-05,
+      "loss": 0.1294,
+      "step": 3794
+    },
+    {
+      "epoch": 31.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1477,
+      "step": 3795
+    },
+    {
+      "epoch": 31.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1443,
+      "step": 3796
+    },
+    {
+      "epoch": 31.12,
+      "learning_rate": 3e-05,
+      "loss": 0.0965,
+      "step": 3797
+    },
+    {
+      "epoch": 31.13,
+      "learning_rate": 3e-05,
+      "loss": 0.1832,
+      "step": 3798
+    },
+    {
+      "epoch": 31.14,
+      "learning_rate": 3e-05,
+      "loss": 0.155,
+      "step": 3799
+    },
+    {
+      "epoch": 31.15,
+      "learning_rate": 3e-05,
+      "loss": 0.1792,
+      "step": 3800
+    },
+    {
+      "epoch": 31.16,
+      "learning_rate": 3e-05,
+      "loss": 0.087,
+      "step": 3801
+    },
+    {
+      "epoch": 31.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1483,
+      "step": 3802
+    },
+    {
+      "epoch": 31.17,
+      "learning_rate": 3e-05,
+      "loss": 0.1925,
+      "step": 3803
+    },
+    {
+      "epoch": 31.18,
+      "learning_rate": 3e-05,
+      "loss": 0.188,
+      "step": 3804
+    },
+    {
+      "epoch": 31.19,
+      "learning_rate": 3e-05,
+      "loss": 0.205,
+      "step": 3805
+    },
+    {
+      "epoch": 31.2,
+      "learning_rate": 3e-05,
+      "loss": 0.2146,
+      "step": 3806
+    },
+    {
+      "epoch": 31.2,
+      "learning_rate": 3e-05,
+      "loss": 0.121,
+      "step": 3807
+    },
+    {
+      "epoch": 31.21,
+      "learning_rate": 3e-05,
+      "loss": 0.1465,
+      "step": 3808
+    },
+    {
+      "epoch": 31.22,
+      "learning_rate": 3e-05,
+      "loss": 0.1702,
+      "step": 3809
+    },
+    {
+      "epoch": 31.23,
+      "learning_rate": 3e-05,
+      "loss": 0.1208,
+      "step": 3810
+    },
+    {
+      "epoch": 31.24,
+      "learning_rate": 3e-05,
+      "loss": 0.1299,
+      "step": 3811
+    },
+    {
+      "epoch": 31.25,
+      "learning_rate": 3e-05,
+      "loss": 0.116,
+      "step": 3812
+    },
+    {
+      "epoch": 31.25,
+      "learning_rate": 3e-05,
+      "loss": 0.141,
+      "step": 3813
+    },
+    {
+      "epoch": 31.26,
+      "learning_rate": 3e-05,
+      "loss": 0.1493,
+      "step": 3814
+    },
+    {
+      "epoch": 31.27,
+      "learning_rate": 3e-05,
+      "loss": 0.1656,
+      "step": 3815
+    },
+    {
+      "epoch": 31.28,
+      "learning_rate": 3e-05,
+      "loss": 0.1074,
+      "step": 3816
+    },
+    {
+      "epoch": 31.29,
+      "learning_rate": 3e-05,
+      "loss": 0.1525,
+      "step": 3817
+    },
+    {
+      "epoch": 31.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1022,
+      "step": 3818
+    },
+    {
+      "epoch": 31.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1481,
+      "step": 3819
+    },
+    {
+      "epoch": 31.31,
+      "learning_rate": 3e-05,
+      "loss": 0.1505,
+      "step": 3820
+    },
+    {
+      "epoch": 31.32,
+      "learning_rate": 3e-05,
+      "loss": 0.1389,
+      "step": 3821
+    },
+    {
+      "epoch": 31.33,
+      "learning_rate": 3e-05,
+      "loss": 0.2499,
+      "step": 3822
+    },
+    {
+      "epoch": 31.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1578,
+      "step": 3823
+    },
+    {
+      "epoch": 31.34,
+      "learning_rate": 3e-05,
+      "loss": 0.2162,
+      "step": 3824
+    },
+    {
+      "epoch": 31.35,
+      "learning_rate": 3e-05,
+      "loss": 0.1497,
+      "step": 3825
+    },
+    {
+      "epoch": 31.36,
+      "learning_rate": 3e-05,
+      "loss": 0.1306,
+      "step": 3826
+    },
+    {
+      "epoch": 31.37,
+      "learning_rate": 3e-05,
+      "loss": 0.1161,
+      "step": 3827
+    },
+    {
+      "epoch": 31.38,
+      "learning_rate": 3e-05,
+      "loss": 0.2471,
+      "step": 3828
+    },
+    {
+      "epoch": 31.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1935,
+      "step": 3829
+    },
+    {
+      "epoch": 31.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1724,
+      "step": 3830
+    },
+    {
+      "epoch": 31.4,
+      "learning_rate": 3e-05,
+      "loss": 0.1877,
+      "step": 3831
+    },
+    {
+      "epoch": 31.41,
+      "learning_rate": 3e-05,
+      "loss": 0.1768,
+      "step": 3832
+    },
+    {
+      "epoch": 31.42,
+      "learning_rate": 3e-05,
+      "loss": 0.1451,
+      "step": 3833
+    },
+    {
+      "epoch": 31.43,
+      "learning_rate": 3e-05,
+      "loss": 0.1721,
+      "step": 3834
+    },
+    {
+      "epoch": 31.43,
+      "learning_rate": 3e-05,
+      "loss": 0.1943,
+      "step": 3835
+    },
+    {
+      "epoch": 31.44,
+      "learning_rate": 3e-05,
+      "loss": 0.1372,
+      "step": 3836
+    },
+    {
+      "epoch": 31.45,
+      "learning_rate": 3e-05,
+      "loss": 0.2001,
+      "step": 3837
+    },
+    {
+      "epoch": 31.46,
+      "learning_rate": 3e-05,
+      "loss": 0.1081,
+      "step": 3838
+    },
+    {
+      "epoch": 31.47,
+      "learning_rate": 3e-05,
+      "loss": 0.1716,
+      "step": 3839
+    },
+    {
+      "epoch": 31.48,
+      "learning_rate": 3e-05,
+      "loss": 0.2059,
+      "step": 3840
+    },
+    {
+      "epoch": 31.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1282,
+      "step": 3841
+    },
+    {
+      "epoch": 31.49,
+      "learning_rate": 3e-05,
+      "loss": 0.1232,
+      "step": 3842
+    },
+    {
+      "epoch": 31.5,
+      "learning_rate": 3e-05,
+      "loss": 0.2156,
+      "step": 3843
+    },
+    {
+      "epoch": 31.51,
+      "learning_rate": 3e-05,
+      "loss": 0.2188,
+      "step": 3844
+    },
+    {
+      "epoch": 31.52,
+      "learning_rate": 3e-05,
+      "loss": 0.14,
+      "step": 3845
+    },
+    {
+      "epoch": 31.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1057,
+      "step": 3846
+    },
+    {
+      "epoch": 31.53,
+      "learning_rate": 3e-05,
+      "loss": 0.2024,
+      "step": 3847
+    },
+    {
+      "epoch": 31.54,
+      "learning_rate": 3e-05,
+      "loss": 0.1556,
+      "step": 3848
+    },
+    {
+      "epoch": 31.55,
+      "learning_rate": 3e-05,
+      "loss": 0.2361,
+      "step": 3849
+    },
+    {
+      "epoch": 31.56,
+      "learning_rate": 3e-05,
+      "loss": 0.1729,
+      "step": 3850
+    },
+    {
+      "epoch": 31.57,
+      "learning_rate": 3e-05,
+      "loss": 0.165,
+      "step": 3851
+    },
+    {
+      "epoch": 31.57,
+      "learning_rate": 3e-05,
+      "loss": 0.1479,
+      "step": 3852
+    },
+    {
+      "epoch": 31.58,
+      "learning_rate": 3e-05,
+      "loss": 0.1117,
+      "step": 3853
+    },
+    {
+      "epoch": 31.59,
+      "learning_rate": 3e-05,
+      "loss": 0.1824,
+      "step": 3854
+    },
+    {
+      "epoch": 31.6,
+      "learning_rate": 3e-05,
+      "loss": 0.1272,
+      "step": 3855
+    },
+    {
+      "epoch": 31.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1209,
+      "step": 3856
+    },
+    {
+      "epoch": 31.61,
+      "learning_rate": 3e-05,
+      "loss": 0.208,
+      "step": 3857
+    },
+    {
+      "epoch": 31.62,
+      "learning_rate": 3e-05,
+      "loss": 0.1581,
+      "step": 3858
+    },
+    {
+      "epoch": 31.63,
+      "learning_rate": 3e-05,
+      "loss": 0.179,
+      "step": 3859
+    },
+    {
+      "epoch": 31.64,
+      "learning_rate": 3e-05,
+      "loss": 0.1749,
+      "step": 3860
+    },
+    {
+      "epoch": 31.65,
+      "learning_rate": 3e-05,
+      "loss": 0.1169,
+      "step": 3861
+    },
+    {
+      "epoch": 31.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1682,
+      "step": 3862
+    },
+    {
+      "epoch": 31.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1802,
+      "step": 3863
+    },
+    {
+      "epoch": 31.67,
+      "learning_rate": 3e-05,
+      "loss": 0.1274,
+      "step": 3864
+    },
+    {
+      "epoch": 31.68,
+      "learning_rate": 3e-05,
+      "loss": 0.1799,
+      "step": 3865
+    },
+    {
+      "epoch": 31.69,
+      "learning_rate": 3e-05,
+      "loss": 0.2332,
+      "step": 3866
+    },
+    {
+      "epoch": 31.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1482,
+      "step": 3867
+    },
+    {
+      "epoch": 31.7,
+      "learning_rate": 3e-05,
+      "loss": 0.158,
+      "step": 3868
+    },
+    {
+      "epoch": 31.71,
+      "learning_rate": 3e-05,
+      "loss": 0.1678,
+      "step": 3869
+    },
+    {
+      "epoch": 31.72,
+      "learning_rate": 3e-05,
+      "loss": 0.2229,
+      "step": 3870
+    },
+    {
+      "epoch": 31.73,
+      "learning_rate": 3e-05,
+      "loss": 0.2214,
+      "step": 3871
+    },
+    {
+      "epoch": 31.74,
+      "learning_rate": 3e-05,
+      "loss": 0.2372,
+      "step": 3872
+    },
+    {
+      "epoch": 31.75,
+      "learning_rate": 3e-05,
+      "loss": 0.2268,
+      "step": 3873
+    },
+    {
+      "epoch": 31.75,
+      "learning_rate": 3e-05,
+      "loss": 0.1506,
+      "step": 3874
+    },
+    {
+      "epoch": 31.76,
+      "learning_rate": 3e-05,
+      "loss": 0.1331,
+      "step": 3875
+    },
+    {
+      "epoch": 31.77,
+      "learning_rate": 3e-05,
+      "loss": 0.2119,
+      "step": 3876
+    },
+    {
+      "epoch": 31.78,
+      "learning_rate": 3e-05,
+      "loss": 0.119,
+      "step": 3877
+    },
+    {
+      "epoch": 31.79,
+      "learning_rate": 3e-05,
+      "loss": 0.1691,
+      "step": 3878
+    },
+    {
+      "epoch": 31.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1914,
+      "step": 3879
+    },
+    {
+      "epoch": 31.8,
+      "learning_rate": 3e-05,
+      "loss": 0.2006,
+      "step": 3880
+    },
+    {
+      "epoch": 31.81,
+      "learning_rate": 3e-05,
+      "loss": 0.1506,
+      "step": 3881
+    },
+    {
+      "epoch": 31.82,
+      "learning_rate": 3e-05,
+      "loss": 0.1578,
+      "step": 3882
+    },
+    {
+      "epoch": 31.83,
+      "learning_rate": 3e-05,
+      "loss": 0.2418,
+      "step": 3883
+    },
+    {
+      "epoch": 31.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1349,
+      "step": 3884
+    },
+    {
+      "epoch": 31.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1184,
+      "step": 3885
+    },
+    {
+      "epoch": 31.85,
+      "learning_rate": 3e-05,
+      "loss": 0.3494,
+      "step": 3886
+    },
+    {
+      "epoch": 31.86,
+      "learning_rate": 3e-05,
+      "loss": 0.1715,
+      "step": 3887
+    },
+    {
+      "epoch": 31.87,
+      "learning_rate": 3e-05,
+      "loss": 0.1973,
+      "step": 3888
+    },
+    {
+      "epoch": 31.88,
+      "learning_rate": 3e-05,
+      "loss": 0.1346,
+      "step": 3889
+    },
+    {
+      "epoch": 31.89,
+      "learning_rate": 3e-05,
+      "loss": 0.204,
+      "step": 3890
+    },
+    {
+      "epoch": 31.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1498,
+      "step": 3891
+    },
+    {
+      "epoch": 31.9,
+      "learning_rate": 3e-05,
+      "loss": 0.1853,
+      "step": 3892
+    },
+    {
+      "epoch": 31.91,
+      "learning_rate": 3e-05,
+      "loss": 0.2186,
+      "step": 3893
+    },
+    {
+      "epoch": 31.92,
+      "learning_rate": 3e-05,
+      "loss": 0.1941,
+      "step": 3894
+    },
+    {
+      "epoch": 31.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1647,
+      "step": 3895
+    },
+    {
+      "epoch": 31.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1644,
+      "step": 3896
+    },
+    {
+      "epoch": 31.94,
+      "learning_rate": 3e-05,
+      "loss": 0.2644,
+      "step": 3897
+    },
+    {
+      "epoch": 31.95,
+      "learning_rate": 3e-05,
+      "loss": 0.1823,
+      "step": 3898
+    },
+    {
+      "epoch": 31.96,
+      "learning_rate": 3e-05,
+      "loss": 0.1015,
+      "step": 3899
+    },
+    {
+      "epoch": 31.97,
+      "learning_rate": 3e-05,
+      "loss": 0.1541,
+      "step": 3900
+    },
+    {
+      "epoch": 31.98,
+      "learning_rate": 3e-05,
+      "loss": 0.198,
+      "step": 3901
+    },
+    {
+      "epoch": 31.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1779,
+      "step": 3902
+    },
+    {
+      "epoch": 31.99,
+      "learning_rate": 3e-05,
+      "loss": 0.207,
+      "step": 3903
+    },
+    {
+      "epoch": 32.0,
+      "learning_rate": 3e-05,
+      "loss": 0.1694,
+      "step": 3904
+    },
+    {
+      "epoch": 32.0,
+      "eval_loss": 0.266357421875,
+      "eval_mae": 0.376953125,
+      "eval_mse": 0.266357421875,
+      "eval_runtime": 51.1571,
+      "eval_samples_per_second": 3233.04,
+      "eval_steps_per_second": 0.274,
+      "step": 3904
+    },
+    {
+      "epoch": 32.01,
+      "learning_rate": 3e-05,
+      "loss": 0.1164,
+      "step": 3905
+    },
+    {
+      "epoch": 32.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1918,
+      "step": 3906
+    },
+    {
+      "epoch": 32.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1698,
+      "step": 3907
+    },
+    {
+      "epoch": 32.03,
+      "learning_rate": 3e-05,
+      "loss": 0.2041,
+      "step": 3908
+    },
+    {
+      "epoch": 32.04,
+      "learning_rate": 3e-05,
+      "loss": 0.1609,
+      "step": 3909
+    },
+    {
+      "epoch": 32.05,
+      "learning_rate": 3e-05,
+      "loss": 0.1414,
+      "step": 3910
+    },
+    {
+      "epoch": 32.06,
+      "learning_rate": 3e-05,
+      "loss": 0.1289,
+      "step": 3911
+    },
+    {
+      "epoch": 32.07,
+      "learning_rate": 3e-05,
+      "loss": 0.2385,
+      "step": 3912
+    },
+    {
+      "epoch": 32.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1803,
+      "step": 3913
+    },
+    {
+      "epoch": 32.08,
+      "learning_rate": 3e-05,
+      "loss": 0.1744,
+      "step": 3914
+    },
+    {
+      "epoch": 32.09,
+      "learning_rate": 3e-05,
+      "loss": 0.1953,
+      "step": 3915
+    },
+    {
+      "epoch": 32.1,
+      "learning_rate": 3e-05,
+      "loss": 0.1599,
+      "step": 3916
+    },
+    {
+      "epoch": 32.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1293,
+      "step": 3917
+    },
+    {
+      "epoch": 32.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1918,
+      "step": 3918
+    },
+    {
+      "epoch": 32.12,
+      "learning_rate": 3e-05,
+      "loss": 0.2185,
+      "step": 3919
+    },
+    {
+      "epoch": 32.13,
+      "learning_rate": 3e-05,
+      "loss": 0.1581,
+      "step": 3920
+    },
+    {
+      "epoch": 32.14,
+      "learning_rate": 3e-05,
+      "loss": 0.1075,
+      "step": 3921
+    },
+    {
+      "epoch": 32.15,
+      "learning_rate": 3e-05,
+      "loss": 0.197,
+      "step": 3922
+    },
+    {
+      "epoch": 32.16,
+      "learning_rate": 3e-05,
+      "loss": 0.2389,
+      "step": 3923
+    },
+    {
+      "epoch": 32.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1621,
+      "step": 3924
+    },
+    {
+      "epoch": 32.17,
+      "learning_rate": 3e-05,
+      "loss": 0.0998,
+      "step": 3925
+    },
+    {
+      "epoch": 32.18,
+      "learning_rate": 3e-05,
+      "loss": 0.1648,
+      "step": 3926
+    },
+    {
+      "epoch": 32.19,
+      "learning_rate": 3e-05,
+      "loss": 0.165,
+      "step": 3927
+    },
+    {
+      "epoch": 32.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1465,
+      "step": 3928
+    },
+    {
+      "epoch": 32.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1399,
+      "step": 3929
+    },
+    {
+      "epoch": 32.21,
+      "learning_rate": 3e-05,
+      "loss": 0.1729,
+      "step": 3930
+    },
+    {
+      "epoch": 32.22,
+      "learning_rate": 3e-05,
+      "loss": 0.165,
+      "step": 3931
+    },
+    {
+      "epoch": 32.23,
+      "learning_rate": 3e-05,
+      "loss": 0.1821,
+      "step": 3932
+    },
+    {
+      "epoch": 32.24,
+      "learning_rate": 3e-05,
+      "loss": 0.1223,
+      "step": 3933
+    },
+    {
+      "epoch": 32.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1283,
+      "step": 3934
+    },
+    {
+      "epoch": 32.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1733,
+      "step": 3935
+    },
+    {
+      "epoch": 32.26,
+      "learning_rate": 3e-05,
+      "loss": 0.1702,
+      "step": 3936
+    },
+    {
+      "epoch": 32.27,
+      "learning_rate": 3e-05,
+      "loss": 0.1624,
+      "step": 3937
+    },
+    {
+      "epoch": 32.28,
+      "learning_rate": 3e-05,
+      "loss": 0.1608,
+      "step": 3938
+    },
+    {
+      "epoch": 32.29,
+      "learning_rate": 3e-05,
+      "loss": 0.1721,
+      "step": 3939
+    },
+    {
+      "epoch": 32.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1354,
+      "step": 3940
+    },
+    {
+      "epoch": 32.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1295,
+      "step": 3941
+    },
+    {
+      "epoch": 32.31,
+      "learning_rate": 3e-05,
+      "loss": 0.1731,
+      "step": 3942
+    },
+    {
+      "epoch": 32.32,
+      "learning_rate": 3e-05,
+      "loss": 0.1503,
+      "step": 3943
+    },
+    {
+      "epoch": 32.33,
+      "learning_rate": 3e-05,
+      "loss": 0.2324,
+      "step": 3944
+    },
+    {
+      "epoch": 32.34,
+      "learning_rate": 3e-05,
+      "loss": 0.2444,
+      "step": 3945
+    },
+    {
+      "epoch": 32.34,
+      "learning_rate": 3e-05,
+      "loss": 0.131,
+      "step": 3946
+    },
+    {
+      "epoch": 32.35,
+      "learning_rate": 3e-05,
+      "loss": 0.1168,
+      "step": 3947
+    },
+    {
+      "epoch": 32.36,
+      "learning_rate": 3e-05,
+      "loss": 0.1127,
+      "step": 3948
+    },
+    {
+      "epoch": 32.37,
+      "learning_rate": 3e-05,
+      "loss": 0.2067,
+      "step": 3949
+    },
+    {
+      "epoch": 32.38,
+      "learning_rate": 3e-05,
+      "loss": 0.1346,
+      "step": 3950
+    },
+    {
+      "epoch": 32.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1156,
+      "step": 3951
+    },
+    {
+      "epoch": 32.39,
+      "learning_rate": 3e-05,
+      "loss": 0.2097,
+      "step": 3952
+    },
+    {
+      "epoch": 32.4,
+      "learning_rate": 3e-05,
+      "loss": 0.1014,
+      "step": 3953
+    },
+    {
+      "epoch": 32.41,
+      "learning_rate": 3e-05,
+      "loss": 0.1263,
+      "step": 3954
+    },
+    {
+      "epoch": 32.42,
+      "learning_rate": 3e-05,
+      "loss": 0.2673,
+      "step": 3955
+    },
+    {
+      "epoch": 32.43,
+      "learning_rate": 3e-05,
+      "loss": 0.1632,
+      "step": 3956
+    },
+    {
+      "epoch": 32.43,
+      "learning_rate": 3e-05,
+      "loss": 0.1304,
+      "step": 3957
+    },
+    {
+      "epoch": 32.44,
+      "learning_rate": 3e-05,
+      "loss": 0.1643,
+      "step": 3958
+    },
+    {
+      "epoch": 32.45,
+      "learning_rate": 3e-05,
+      "loss": 0.1298,
+      "step": 3959
+    },
+    {
+      "epoch": 32.46,
+      "learning_rate": 3e-05,
+      "loss": 0.178,
+      "step": 3960
+    },
+    {
+      "epoch": 32.47,
+      "learning_rate": 3e-05,
+      "loss": 0.1907,
+      "step": 3961
+    },
+    {
+      "epoch": 32.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1854,
+      "step": 3962
+    },
+    {
+      "epoch": 32.48,
+      "learning_rate": 3e-05,
+      "loss": 0.3452,
+      "step": 3963
+    },
+    {
+      "epoch": 32.49,
+      "learning_rate": 3e-05,
+      "loss": 0.2466,
+      "step": 3964
+    },
+    {
+      "epoch": 32.5,
+      "learning_rate": 3e-05,
+      "loss": 0.1597,
+      "step": 3965
+    },
+    {
+      "epoch": 32.51,
+      "learning_rate": 3e-05,
+      "loss": 0.127,
+      "step": 3966
+    },
+    {
+      "epoch": 32.52,
+      "learning_rate": 3e-05,
+      "loss": 0.156,
+      "step": 3967
+    },
+    {
+      "epoch": 32.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1969,
+      "step": 3968
+    },
+    {
+      "epoch": 32.53,
+      "learning_rate": 3e-05,
+      "loss": 0.202,
+      "step": 3969
+    },
+    {
+      "epoch": 32.54,
+      "learning_rate": 3e-05,
+      "loss": 0.0865,
+      "step": 3970
+    },
+    {
+      "epoch": 32.55,
+      "learning_rate": 3e-05,
+      "loss": 0.1888,
+      "step": 3971
+    },
+    {
+      "epoch": 32.56,
+      "learning_rate": 3e-05,
+      "loss": 0.1371,
+      "step": 3972
+    },
+    {
+      "epoch": 32.57,
+      "learning_rate": 3e-05,
+      "loss": 0.1805,
+      "step": 3973
+    },
+    {
+      "epoch": 32.57,
+      "learning_rate": 3e-05,
+      "loss": 0.2964,
+      "step": 3974
+    },
+    {
+      "epoch": 32.58,
+      "learning_rate": 3e-05,
+      "loss": 0.1147,
+      "step": 3975
+    },
+    {
+      "epoch": 32.59,
+      "learning_rate": 3e-05,
+      "loss": 0.2502,
+      "step": 3976
+    },
+    {
+      "epoch": 32.6,
+      "learning_rate": 3e-05,
+      "loss": 0.1481,
+      "step": 3977
+    },
+    {
+      "epoch": 32.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1993,
+      "step": 3978
+    },
+    {
+      "epoch": 32.61,
+      "learning_rate": 3e-05,
+      "loss": 0.2607,
+      "step": 3979
+    },
+    {
+      "epoch": 32.62,
+      "learning_rate": 3e-05,
+      "loss": 0.1384,
+      "step": 3980
+    },
+    {
+      "epoch": 32.63,
+      "learning_rate": 3e-05,
+      "loss": 0.1442,
+      "step": 3981
+    },
+    {
+      "epoch": 32.64,
+      "learning_rate": 3e-05,
+      "loss": 0.1298,
+      "step": 3982
+    },
+    {
+      "epoch": 32.65,
+      "learning_rate": 3e-05,
+      "loss": 0.1573,
+      "step": 3983
+    },
+    {
+      "epoch": 32.66,
+      "learning_rate": 3e-05,
+      "loss": 0.2135,
+      "step": 3984
+    },
+    {
+      "epoch": 32.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1437,
+      "step": 3985
+    },
+    {
+      "epoch": 32.67,
+      "learning_rate": 3e-05,
+      "loss": 0.2164,
+      "step": 3986
+    },
+    {
+      "epoch": 32.68,
+      "learning_rate": 3e-05,
+      "loss": 0.2198,
+      "step": 3987
+    },
+    {
+      "epoch": 32.69,
+      "learning_rate": 3e-05,
+      "loss": 0.1208,
+      "step": 3988
+    },
+    {
+      "epoch": 32.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1422,
+      "step": 3989
+    },
+    {
+      "epoch": 32.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1854,
+      "step": 3990
+    },
+    {
+      "epoch": 32.71,
+      "learning_rate": 3e-05,
+      "loss": 0.1982,
+      "step": 3991
+    },
+    {
+      "epoch": 32.72,
+      "learning_rate": 3e-05,
+      "loss": 0.1658,
+      "step": 3992
+    },
+    {
+      "epoch": 32.73,
+      "learning_rate": 3e-05,
+      "loss": 0.1615,
+      "step": 3993
+    },
+    {
+      "epoch": 32.74,
+      "learning_rate": 3e-05,
+      "loss": 0.1691,
+      "step": 3994
+    },
+    {
+      "epoch": 32.75,
+      "learning_rate": 3e-05,
+      "loss": 0.3376,
+      "step": 3995
+    },
+    {
+      "epoch": 32.75,
+      "learning_rate": 3e-05,
+      "loss": 0.1421,
+      "step": 3996
+    },
+    {
+      "epoch": 32.76,
+      "learning_rate": 3e-05,
+      "loss": 0.1698,
+      "step": 3997
+    },
+    {
+      "epoch": 32.77,
+      "learning_rate": 3e-05,
+      "loss": 0.1648,
+      "step": 3998
+    },
+    {
+      "epoch": 32.78,
+      "learning_rate": 3e-05,
+      "loss": 0.173,
+      "step": 3999
+    },
+    {
+      "epoch": 32.79,
+      "learning_rate": 3e-05,
+      "loss": 0.1887,
+      "step": 4000
+    },
+    {
+      "epoch": 32.8,
+      "learning_rate": 3e-05,
+      "loss": 0.2057,
+      "step": 4001
+    },
+    {
+      "epoch": 32.8,
+      "learning_rate": 3e-05,
+      "loss": 0.243,
+      "step": 4002
+    },
+    {
+      "epoch": 32.81,
+      "learning_rate": 3e-05,
+      "loss": 0.1705,
+      "step": 4003
+    },
+    {
+      "epoch": 32.82,
+      "learning_rate": 3e-05,
+      "loss": 0.182,
+      "step": 4004
+    },
+    {
+      "epoch": 32.83,
+      "learning_rate": 3e-05,
+      "loss": 0.1941,
+      "step": 4005
+    },
+    {
+      "epoch": 32.84,
+      "learning_rate": 3e-05,
+      "loss": 0.2104,
+      "step": 4006
+    },
+    {
+      "epoch": 32.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1547,
+      "step": 4007
+    },
+    {
+      "epoch": 32.85,
+      "learning_rate": 3e-05,
+      "loss": 0.175,
+      "step": 4008
+    },
+    {
+      "epoch": 32.86,
+      "learning_rate": 3e-05,
+      "loss": 0.1049,
+      "step": 4009
+    },
+    {
+      "epoch": 32.87,
+      "learning_rate": 3e-05,
+      "loss": 0.1749,
+      "step": 4010
+    },
+    {
+      "epoch": 32.88,
+      "learning_rate": 3e-05,
+      "loss": 0.1384,
+      "step": 4011
+    },
+    {
+      "epoch": 32.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1443,
+      "step": 4012
+    },
+    {
+      "epoch": 32.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1202,
+      "step": 4013
+    },
+    {
+      "epoch": 32.9,
+      "learning_rate": 3e-05,
+      "loss": 0.1378,
+      "step": 4014
+    },
+    {
+      "epoch": 32.91,
+      "learning_rate": 3e-05,
+      "loss": 0.1874,
+      "step": 4015
+    },
+    {
+      "epoch": 32.92,
+      "learning_rate": 3e-05,
+      "loss": 0.2166,
+      "step": 4016
+    },
+    {
+      "epoch": 32.93,
+      "learning_rate": 3e-05,
+      "loss": 0.0878,
+      "step": 4017
+    },
+    {
+      "epoch": 32.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1361,
+      "step": 4018
+    },
+    {
+      "epoch": 32.94,
+      "learning_rate": 3e-05,
+      "loss": 0.1606,
+      "step": 4019
+    },
+    {
+      "epoch": 32.95,
+      "learning_rate": 3e-05,
+      "loss": 0.1768,
+      "step": 4020
+    },
+    {
+      "epoch": 32.96,
+      "learning_rate": 3e-05,
+      "loss": 0.1105,
+      "step": 4021
+    },
+    {
+      "epoch": 32.97,
+      "learning_rate": 3e-05,
+      "loss": 0.2174,
+      "step": 4022
+    },
+    {
+      "epoch": 32.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1423,
+      "step": 4023
+    },
+    {
+      "epoch": 32.98,
+      "learning_rate": 3e-05,
+      "loss": 0.2188,
+      "step": 4024
+    },
+    {
+      "epoch": 32.99,
+      "learning_rate": 3e-05,
+      "loss": 0.1539,
+      "step": 4025
+    },
+    {
+      "epoch": 33.0,
+      "learning_rate": 3e-05,
+      "loss": 0.2169,
+      "step": 4026
+    },
+    {
+      "epoch": 33.0,
+      "eval_loss": 0.265869140625,
+      "eval_mae": 0.375244140625,
+      "eval_mse": 0.265869140625,
+      "eval_runtime": 51.1016,
+      "eval_samples_per_second": 3236.551,
+      "eval_steps_per_second": 0.274,
+      "step": 4026
+    },
+    {
+      "epoch": 33.01,
+      "learning_rate": 3e-05,
+      "loss": 0.099,
+      "step": 4027
+    },
+    {
+      "epoch": 33.02,
+      "learning_rate": 3e-05,
+      "loss": 0.2029,
+      "step": 4028
+    },
+    {
+      "epoch": 33.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1072,
+      "step": 4029
+    },
+    {
+      "epoch": 33.03,
+      "learning_rate": 3e-05,
+      "loss": 0.2312,
+      "step": 4030
+    },
+    {
+      "epoch": 33.04,
+      "learning_rate": 3e-05,
+      "loss": 0.1439,
+      "step": 4031
+    },
+    {
+      "epoch": 33.05,
+      "learning_rate": 3e-05,
+      "loss": 0.1261,
+      "step": 4032
+    },
+    {
+      "epoch": 33.06,
+      "learning_rate": 3e-05,
+      "loss": 0.1609,
+      "step": 4033
+    },
+    {
+      "epoch": 33.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1052,
+      "step": 4034
+    },
+    {
+      "epoch": 33.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1783,
+      "step": 4035
+    },
+    {
+      "epoch": 33.08,
+      "learning_rate": 3e-05,
+      "loss": 0.2047,
+      "step": 4036
+    },
+    {
+      "epoch": 33.09,
+      "learning_rate": 3e-05,
+      "loss": 0.1501,
+      "step": 4037
+    },
+    {
+      "epoch": 33.1,
+      "learning_rate": 3e-05,
+      "loss": 0.2661,
+      "step": 4038
+    },
+    {
+      "epoch": 33.11,
+      "learning_rate": 3e-05,
+      "loss": 0.2078,
+      "step": 4039
+    },
+    {
+      "epoch": 33.11,
+      "learning_rate": 3e-05,
+      "loss": 0.15,
+      "step": 4040
+    },
+    {
+      "epoch": 33.12,
+      "learning_rate": 3e-05,
+      "loss": 0.1956,
+      "step": 4041
+    },
+    {
+      "epoch": 33.13,
+      "learning_rate": 3e-05,
+      "loss": 0.1179,
+      "step": 4042
+    },
+    {
+      "epoch": 33.14,
+      "learning_rate": 3e-05,
+      "loss": 0.1971,
+      "step": 4043
+    },
+    {
+      "epoch": 33.15,
+      "learning_rate": 3e-05,
+      "loss": 0.1348,
+      "step": 4044
+    },
+    {
+      "epoch": 33.16,
+      "learning_rate": 3e-05,
+      "loss": 0.0989,
+      "step": 4045
+    },
+    {
+      "epoch": 33.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1549,
+      "step": 4046
+    },
+    {
+      "epoch": 33.17,
+      "learning_rate": 3e-05,
+      "loss": 0.1066,
+      "step": 4047
+    },
+    {
+      "epoch": 33.18,
+      "learning_rate": 3e-05,
+      "loss": 0.1272,
+      "step": 4048
+    },
+    {
+      "epoch": 33.19,
+      "learning_rate": 3e-05,
+      "loss": 0.1433,
+      "step": 4049
+    },
+    {
+      "epoch": 33.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1335,
+      "step": 4050
+    },
+    {
+      "epoch": 33.2,
+      "learning_rate": 3e-05,
+      "loss": 0.3723,
+      "step": 4051
+    },
+    {
+      "epoch": 33.21,
+      "learning_rate": 3e-05,
+      "loss": 0.1274,
+      "step": 4052
+    },
+    {
+      "epoch": 33.22,
+      "learning_rate": 3e-05,
+      "loss": 0.1578,
+      "step": 4053
+    },
+    {
+      "epoch": 33.23,
+      "learning_rate": 3e-05,
+      "loss": 0.1646,
+      "step": 4054
+    },
+    {
+      "epoch": 33.24,
+      "learning_rate": 3e-05,
+      "loss": 0.183,
+      "step": 4055
+    },
+    {
+      "epoch": 33.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1554,
+      "step": 4056
+    },
+    {
+      "epoch": 33.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1724,
+      "step": 4057
+    },
+    {
+      "epoch": 33.26,
+      "learning_rate": 3e-05,
+      "loss": 0.1466,
+      "step": 4058
+    },
+    {
+      "epoch": 33.27,
+      "learning_rate": 3e-05,
+      "loss": 0.163,
+      "step": 4059
+    },
+    {
+      "epoch": 33.28,
+      "learning_rate": 3e-05,
+      "loss": 0.1953,
+      "step": 4060
+    },
+    {
+      "epoch": 33.29,
+      "learning_rate": 3e-05,
+      "loss": 0.191,
+      "step": 4061
+    },
+    {
+      "epoch": 33.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1763,
+      "step": 4062
+    },
+    {
+      "epoch": 33.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1373,
+      "step": 4063
+    },
+    {
+      "epoch": 33.31,
+      "learning_rate": 3e-05,
+      "loss": 0.1265,
+      "step": 4064
+    },
+    {
+      "epoch": 33.32,
+      "learning_rate": 3e-05,
+      "loss": 0.1866,
+      "step": 4065
+    },
+    {
+      "epoch": 33.33,
+      "learning_rate": 3e-05,
+      "loss": 0.1985,
+      "step": 4066
+    },
+    {
+      "epoch": 33.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1149,
+      "step": 4067
+    },
+    {
+      "epoch": 33.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1299,
+      "step": 4068
+    },
+    {
+      "epoch": 33.35,
+      "learning_rate": 3e-05,
+      "loss": 0.1624,
+      "step": 4069
+    },
+    {
+      "epoch": 33.36,
+      "learning_rate": 3e-05,
+      "loss": 0.2218,
+      "step": 4070
+    },
+    {
+      "epoch": 33.37,
+      "learning_rate": 3e-05,
+      "loss": 0.168,
+      "step": 4071
+    },
+    {
+      "epoch": 33.38,
+      "learning_rate": 3e-05,
+      "loss": 0.1207,
+      "step": 4072
+    },
+    {
+      "epoch": 33.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1202,
+      "step": 4073
+    },
+    {
+      "epoch": 33.39,
+      "learning_rate": 3e-05,
+      "loss": 0.163,
+      "step": 4074
+    },
+    {
+      "epoch": 33.4,
+      "learning_rate": 3e-05,
+      "loss": 0.1628,
+      "step": 4075
+    },
+    {
+      "epoch": 33.41,
+      "learning_rate": 3e-05,
+      "loss": 0.1932,
+      "step": 4076
+    },
+    {
+      "epoch": 33.42,
+      "learning_rate": 3e-05,
+      "loss": 0.1062,
+      "step": 4077
+    },
+    {
+      "epoch": 33.43,
+      "learning_rate": 3e-05,
+      "loss": 0.1083,
+      "step": 4078
+    },
+    {
+      "epoch": 33.43,
+      "learning_rate": 3e-05,
+      "loss": 0.2424,
+      "step": 4079
+    },
+    {
+      "epoch": 33.44,
+      "learning_rate": 3e-05,
+      "loss": 0.1394,
+      "step": 4080
+    },
+    {
+      "epoch": 33.45,
+      "learning_rate": 3e-05,
+      "loss": 0.2051,
+      "step": 4081
+    },
+    {
+      "epoch": 33.46,
+      "learning_rate": 3e-05,
+      "loss": 0.1304,
+      "step": 4082
+    },
+    {
+      "epoch": 33.47,
+      "learning_rate": 3e-05,
+      "loss": 0.1465,
+      "step": 4083
+    },
+    {
+      "epoch": 33.48,
+      "learning_rate": 3e-05,
+      "loss": 0.2148,
+      "step": 4084
+    },
+    {
+      "epoch": 33.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1083,
+      "step": 4085
+    },
+    {
+      "epoch": 33.49,
+      "learning_rate": 3e-05,
+      "loss": 0.1975,
+      "step": 4086
+    },
+    {
+      "epoch": 33.5,
+      "learning_rate": 3e-05,
+      "loss": 0.2021,
+      "step": 4087
+    },
+    {
+      "epoch": 33.51,
+      "learning_rate": 3e-05,
+      "loss": 0.1544,
+      "step": 4088
+    },
+    {
+      "epoch": 33.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1262,
+      "step": 4089
+    },
+    {
+      "epoch": 33.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1469,
+      "step": 4090
+    },
+    {
+      "epoch": 33.53,
+      "learning_rate": 3e-05,
+      "loss": 0.1727,
+      "step": 4091
+    },
+    {
+      "epoch": 33.54,
+      "learning_rate": 3e-05,
+      "loss": 0.1879,
+      "step": 4092
+    },
+    {
+      "epoch": 33.55,
+      "learning_rate": 3e-05,
+      "loss": 0.1658,
+      "step": 4093
+    },
+    {
+      "epoch": 33.56,
+      "learning_rate": 3e-05,
+      "loss": 0.2402,
+      "step": 4094
+    },
+    {
+      "epoch": 33.57,
+      "learning_rate": 3e-05,
+      "loss": 0.2079,
+      "step": 4095
+    },
+    {
+      "epoch": 33.57,
+      "learning_rate": 3e-05,
+      "loss": 0.1003,
+      "step": 4096
+    },
+    {
+      "epoch": 33.58,
+      "learning_rate": 3e-05,
+      "loss": 0.1061,
+      "step": 4097
+    },
+    {
+      "epoch": 33.59,
+      "learning_rate": 3e-05,
+      "loss": 0.1996,
+      "step": 4098
+    },
+    {
+      "epoch": 33.6,
+      "learning_rate": 3e-05,
+      "loss": 0.1342,
+      "step": 4099
+    },
+    {
+      "epoch": 33.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1816,
+      "step": 4100
+    },
+    {
+      "epoch": 33.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1888,
+      "step": 4101
+    },
+    {
+      "epoch": 33.62,
+      "learning_rate": 3e-05,
+      "loss": 0.1247,
+      "step": 4102
+    },
+    {
+      "epoch": 33.63,
+      "learning_rate": 3e-05,
+      "loss": 0.1765,
+      "step": 4103
+    },
+    {
+      "epoch": 33.64,
+      "learning_rate": 3e-05,
+      "loss": 0.1328,
+      "step": 4104
+    },
+    {
+      "epoch": 33.65,
+      "learning_rate": 3e-05,
+      "loss": 0.1517,
+      "step": 4105
+    },
+    {
+      "epoch": 33.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1879,
+      "step": 4106
+    },
+    {
+      "epoch": 33.66,
+      "learning_rate": 3e-05,
+      "loss": 0.2581,
+      "step": 4107
+    },
+    {
+      "epoch": 33.67,
+      "learning_rate": 3e-05,
+      "loss": 0.1549,
+      "step": 4108
+    },
+    {
+      "epoch": 33.68,
+      "learning_rate": 3e-05,
+      "loss": 0.1555,
+      "step": 4109
+    },
+    {
+      "epoch": 33.69,
+      "learning_rate": 3e-05,
+      "loss": 0.2347,
+      "step": 4110
+    },
+    {
+      "epoch": 33.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1783,
+      "step": 4111
+    },
+    {
+      "epoch": 33.7,
+      "learning_rate": 3e-05,
+      "loss": 0.2191,
+      "step": 4112
+    },
+    {
+      "epoch": 33.71,
+      "learning_rate": 3e-05,
+      "loss": 0.1589,
+      "step": 4113
+    },
+    {
+      "epoch": 33.72,
+      "learning_rate": 3e-05,
+      "loss": 0.2294,
+      "step": 4114
+    },
+    {
+      "epoch": 33.73,
+      "learning_rate": 3e-05,
+      "loss": 0.1337,
+      "step": 4115
+    },
+    {
+      "epoch": 33.74,
+      "learning_rate": 3e-05,
+      "loss": 0.1584,
+      "step": 4116
+    },
+    {
+      "epoch": 33.75,
+      "learning_rate": 3e-05,
+      "loss": 0.1527,
+      "step": 4117
+    },
+    {
+      "epoch": 33.75,
+      "learning_rate": 3e-05,
+      "loss": 0.194,
+      "step": 4118
+    },
+    {
+      "epoch": 33.76,
+      "learning_rate": 3e-05,
+      "loss": 0.1449,
+      "step": 4119
+    },
+    {
+      "epoch": 33.77,
+      "learning_rate": 3e-05,
+      "loss": 0.1674,
+      "step": 4120
+    },
+    {
+      "epoch": 33.78,
+      "learning_rate": 3e-05,
+      "loss": 0.1185,
+      "step": 4121
+    },
+    {
+      "epoch": 33.79,
+      "learning_rate": 3e-05,
+      "loss": 0.2712,
+      "step": 4122
+    },
+    {
+      "epoch": 33.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1294,
+      "step": 4123
+    },
+    {
+      "epoch": 33.8,
+      "learning_rate": 3e-05,
+      "loss": 0.2228,
+      "step": 4124
+    },
+    {
+      "epoch": 33.81,
+      "learning_rate": 3e-05,
+      "loss": 0.1716,
+      "step": 4125
+    },
+    {
+      "epoch": 33.82,
+      "learning_rate": 3e-05,
+      "loss": 0.1515,
+      "step": 4126
+    },
+    {
+      "epoch": 33.83,
+      "learning_rate": 3e-05,
+      "loss": 0.1093,
+      "step": 4127
+    },
+    {
+      "epoch": 33.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1132,
+      "step": 4128
+    },
+    {
+      "epoch": 33.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1237,
+      "step": 4129
+    },
+    {
+      "epoch": 33.85,
+      "learning_rate": 3e-05,
+      "loss": 0.1079,
+      "step": 4130
+    },
+    {
+      "epoch": 33.86,
+      "learning_rate": 3e-05,
+      "loss": 0.1859,
+      "step": 4131
+    },
+    {
+      "epoch": 33.87,
+      "learning_rate": 3e-05,
+      "loss": 0.2312,
+      "step": 4132
+    },
+    {
+      "epoch": 33.88,
+      "learning_rate": 3e-05,
+      "loss": 0.1826,
+      "step": 4133
+    },
+    {
+      "epoch": 33.89,
+      "learning_rate": 3e-05,
+      "loss": 0.2493,
+      "step": 4134
+    },
+    {
+      "epoch": 33.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1508,
+      "step": 4135
+    },
+    {
+      "epoch": 33.9,
+      "learning_rate": 3e-05,
+      "loss": 0.2345,
+      "step": 4136
+    },
+    {
+      "epoch": 33.91,
+      "learning_rate": 3e-05,
+      "loss": 0.2421,
+      "step": 4137
+    },
+    {
+      "epoch": 33.92,
+      "learning_rate": 3e-05,
+      "loss": 0.1871,
+      "step": 4138
+    },
+    {
+      "epoch": 33.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1787,
+      "step": 4139
+    },
+    {
+      "epoch": 33.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1998,
+      "step": 4140
+    },
+    {
+      "epoch": 33.94,
+      "learning_rate": 3e-05,
+      "loss": 0.2112,
+      "step": 4141
+    },
+    {
+      "epoch": 33.95,
+      "learning_rate": 3e-05,
+      "loss": 0.1901,
+      "step": 4142
+    },
+    {
+      "epoch": 33.96,
+      "learning_rate": 3e-05,
+      "loss": 0.1377,
+      "step": 4143
+    },
+    {
+      "epoch": 33.97,
+      "learning_rate": 3e-05,
+      "loss": 0.2457,
+      "step": 4144
+    },
+    {
+      "epoch": 33.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1371,
+      "step": 4145
+    },
+    {
+      "epoch": 33.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1337,
+      "step": 4146
+    },
+    {
+      "epoch": 33.99,
+      "learning_rate": 3e-05,
+      "loss": 0.124,
+      "step": 4147
+    },
+    {
+      "epoch": 34.0,
+      "learning_rate": 3e-05,
+      "loss": 0.4246,
+      "step": 4148
+    },
+    {
+      "epoch": 34.0,
+      "eval_loss": 0.2666015625,
+      "eval_mae": 0.37548828125,
+      "eval_mse": 0.2666015625,
+      "eval_runtime": 51.1345,
+      "eval_samples_per_second": 3234.472,
+      "eval_steps_per_second": 0.274,
+      "step": 4148
+    },
+    {
+      "epoch": 34.01,
+      "learning_rate": 3e-05,
+      "loss": 0.2974,
+      "step": 4149
+    },
+    {
+      "epoch": 34.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1735,
+      "step": 4150
+    },
+    {
+      "epoch": 34.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1565,
+      "step": 4151
+    },
+    {
+      "epoch": 34.03,
+      "learning_rate": 3e-05,
+      "loss": 0.113,
+      "step": 4152
+    },
+    {
+      "epoch": 34.04,
+      "learning_rate": 3e-05,
+      "loss": 0.156,
+      "step": 4153
+    },
+    {
+      "epoch": 34.05,
+      "learning_rate": 3e-05,
+      "loss": 0.1361,
+      "step": 4154
+    },
+    {
+      "epoch": 34.06,
+      "learning_rate": 3e-05,
+      "loss": 0.1622,
+      "step": 4155
+    },
+    {
+      "epoch": 34.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1455,
+      "step": 4156
+    },
+    {
+      "epoch": 34.07,
+      "learning_rate": 3e-05,
+      "loss": 0.2034,
+      "step": 4157
+    },
+    {
+      "epoch": 34.08,
+      "learning_rate": 3e-05,
+      "loss": 0.15,
+      "step": 4158
+    },
+    {
+      "epoch": 34.09,
+      "learning_rate": 3e-05,
+      "loss": 0.1902,
+      "step": 4159
+    },
+    {
+      "epoch": 34.1,
+      "learning_rate": 3e-05,
+      "loss": 0.2135,
+      "step": 4160
+    },
+    {
+      "epoch": 34.11,
+      "learning_rate": 3e-05,
+      "loss": 0.2273,
+      "step": 4161
+    },
+    {
+      "epoch": 34.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1647,
+      "step": 4162
+    },
+    {
+      "epoch": 34.12,
+      "learning_rate": 3e-05,
+      "loss": 0.1746,
+      "step": 4163
+    },
+    {
+      "epoch": 34.13,
+      "learning_rate": 3e-05,
+      "loss": 0.1903,
+      "step": 4164
+    },
+    {
+      "epoch": 34.14,
+      "learning_rate": 3e-05,
+      "loss": 0.1179,
+      "step": 4165
+    },
+    {
+      "epoch": 34.15,
+      "learning_rate": 3e-05,
+      "loss": 0.1663,
+      "step": 4166
+    },
+    {
+      "epoch": 34.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1389,
+      "step": 4167
+    },
+    {
+      "epoch": 34.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1936,
+      "step": 4168
+    },
+    {
+      "epoch": 34.17,
+      "learning_rate": 3e-05,
+      "loss": 0.1617,
+      "step": 4169
+    },
+    {
+      "epoch": 34.18,
+      "learning_rate": 3e-05,
+      "loss": 0.2134,
+      "step": 4170
+    },
+    {
+      "epoch": 34.19,
+      "learning_rate": 3e-05,
+      "loss": 0.1945,
+      "step": 4171
+    },
+    {
+      "epoch": 34.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1138,
+      "step": 4172
+    },
+    {
+      "epoch": 34.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1309,
+      "step": 4173
+    },
+    {
+      "epoch": 34.21,
+      "learning_rate": 3e-05,
+      "loss": 0.1658,
+      "step": 4174
+    },
+    {
+      "epoch": 34.22,
+      "learning_rate": 3e-05,
+      "loss": 0.1094,
+      "step": 4175
+    },
+    {
+      "epoch": 34.23,
+      "learning_rate": 3e-05,
+      "loss": 0.1163,
+      "step": 4176
+    },
+    {
+      "epoch": 34.24,
+      "learning_rate": 3e-05,
+      "loss": 0.1312,
+      "step": 4177
+    },
+    {
+      "epoch": 34.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1588,
+      "step": 4178
+    },
+    {
+      "epoch": 34.25,
+      "learning_rate": 3e-05,
+      "loss": 0.139,
+      "step": 4179
+    },
+    {
+      "epoch": 34.26,
+      "learning_rate": 3e-05,
+      "loss": 0.1145,
+      "step": 4180
+    },
+    {
+      "epoch": 34.27,
+      "learning_rate": 3e-05,
+      "loss": 0.1311,
+      "step": 4181
+    },
+    {
+      "epoch": 34.28,
+      "learning_rate": 3e-05,
+      "loss": 0.1793,
+      "step": 4182
+    },
+    {
+      "epoch": 34.29,
+      "learning_rate": 3e-05,
+      "loss": 0.1692,
+      "step": 4183
+    },
+    {
+      "epoch": 34.3,
+      "learning_rate": 3e-05,
+      "loss": 0.141,
+      "step": 4184
+    },
+    {
+      "epoch": 34.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1241,
+      "step": 4185
+    },
+    {
+      "epoch": 34.31,
+      "learning_rate": 3e-05,
+      "loss": 0.1376,
+      "step": 4186
+    },
+    {
+      "epoch": 34.32,
+      "learning_rate": 3e-05,
+      "loss": 0.2054,
+      "step": 4187
+    },
+    {
+      "epoch": 34.33,
+      "learning_rate": 3e-05,
+      "loss": 0.1158,
+      "step": 4188
+    },
+    {
+      "epoch": 34.34,
+      "learning_rate": 3e-05,
+      "loss": 0.151,
+      "step": 4189
+    },
+    {
+      "epoch": 34.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1802,
+      "step": 4190
+    },
+    {
+      "epoch": 34.35,
+      "learning_rate": 3e-05,
+      "loss": 0.1737,
+      "step": 4191
+    },
+    {
+      "epoch": 34.36,
+      "learning_rate": 3e-05,
+      "loss": 0.1511,
+      "step": 4192
+    },
+    {
+      "epoch": 34.37,
+      "learning_rate": 3e-05,
+      "loss": 0.155,
+      "step": 4193
+    },
+    {
+      "epoch": 34.38,
+      "learning_rate": 3e-05,
+      "loss": 0.1415,
+      "step": 4194
+    },
+    {
+      "epoch": 34.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1959,
+      "step": 4195
+    },
+    {
+      "epoch": 34.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1686,
+      "step": 4196
+    },
+    {
+      "epoch": 34.4,
+      "learning_rate": 3e-05,
+      "loss": 0.1611,
+      "step": 4197
+    },
+    {
+      "epoch": 34.41,
+      "learning_rate": 3e-05,
+      "loss": 0.1428,
+      "step": 4198
+    },
+    {
+      "epoch": 34.42,
+      "learning_rate": 3e-05,
+      "loss": 0.1721,
+      "step": 4199
+    },
+    {
+      "epoch": 34.43,
+      "learning_rate": 3e-05,
+      "loss": 0.1495,
+      "step": 4200
+    },
+    {
+      "epoch": 34.43,
+      "learning_rate": 3e-05,
+      "loss": 0.2253,
+      "step": 4201
+    },
+    {
+      "epoch": 34.44,
+      "learning_rate": 3e-05,
+      "loss": 0.1465,
+      "step": 4202
+    },
+    {
+      "epoch": 34.45,
+      "learning_rate": 3e-05,
+      "loss": 0.3157,
+      "step": 4203
+    },
+    {
+      "epoch": 34.46,
+      "learning_rate": 3e-05,
+      "loss": 0.1254,
+      "step": 4204
+    },
+    {
+      "epoch": 34.47,
+      "learning_rate": 3e-05,
+      "loss": 0.1346,
+      "step": 4205
+    },
+    {
+      "epoch": 34.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1555,
+      "step": 4206
+    },
+    {
+      "epoch": 34.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1724,
+      "step": 4207
+    },
+    {
+      "epoch": 34.49,
+      "learning_rate": 3e-05,
+      "loss": 0.1508,
+      "step": 4208
+    },
+    {
+      "epoch": 34.5,
+      "learning_rate": 3e-05,
+      "loss": 0.1266,
+      "step": 4209
+    },
+    {
+      "epoch": 34.51,
+      "learning_rate": 3e-05,
+      "loss": 0.208,
+      "step": 4210
+    },
+    {
+      "epoch": 34.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1732,
+      "step": 4211
+    },
+    {
+      "epoch": 34.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1489,
+      "step": 4212
+    },
+    {
+      "epoch": 34.53,
+      "learning_rate": 3e-05,
+      "loss": 0.116,
+      "step": 4213
+    },
+    {
+      "epoch": 34.54,
+      "learning_rate": 3e-05,
+      "loss": 0.1075,
+      "step": 4214
+    },
+    {
+      "epoch": 34.55,
+      "learning_rate": 3e-05,
+      "loss": 0.1138,
+      "step": 4215
+    },
+    {
+      "epoch": 34.56,
+      "learning_rate": 3e-05,
+      "loss": 0.1744,
+      "step": 4216
+    },
+    {
+      "epoch": 34.57,
+      "learning_rate": 3e-05,
+      "loss": 0.1582,
+      "step": 4217
+    },
+    {
+      "epoch": 34.57,
+      "learning_rate": 3e-05,
+      "loss": 0.1501,
+      "step": 4218
+    },
+    {
+      "epoch": 34.58,
+      "learning_rate": 3e-05,
+      "loss": 0.1548,
+      "step": 4219
+    },
+    {
+      "epoch": 34.59,
+      "learning_rate": 3e-05,
+      "loss": 0.2417,
+      "step": 4220
+    },
+    {
+      "epoch": 34.6,
+      "learning_rate": 3e-05,
+      "loss": 0.2332,
+      "step": 4221
+    },
+    {
+      "epoch": 34.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1473,
+      "step": 4222
+    },
+    {
+      "epoch": 34.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1647,
+      "step": 4223
+    },
+    {
+      "epoch": 34.62,
+      "learning_rate": 3e-05,
+      "loss": 0.1313,
+      "step": 4224
+    },
+    {
+      "epoch": 34.63,
+      "learning_rate": 3e-05,
+      "loss": 0.1267,
+      "step": 4225
+    },
+    {
+      "epoch": 34.64,
+      "learning_rate": 3e-05,
+      "loss": 0.1283,
+      "step": 4226
+    },
+    {
+      "epoch": 34.65,
+      "learning_rate": 3e-05,
+      "loss": 0.2163,
+      "step": 4227
+    },
+    {
+      "epoch": 34.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1281,
+      "step": 4228
+    },
+    {
+      "epoch": 34.66,
+      "learning_rate": 3e-05,
+      "loss": 0.132,
+      "step": 4229
+    },
+    {
+      "epoch": 34.67,
+      "learning_rate": 3e-05,
+      "loss": 0.1004,
+      "step": 4230
+    },
+    {
+      "epoch": 34.68,
+      "learning_rate": 3e-05,
+      "loss": 0.1008,
+      "step": 4231
+    },
+    {
+      "epoch": 34.69,
+      "learning_rate": 3e-05,
+      "loss": 0.0897,
+      "step": 4232
+    },
+    {
+      "epoch": 34.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1267,
+      "step": 4233
+    },
+    {
+      "epoch": 34.7,
+      "learning_rate": 3e-05,
+      "loss": 0.2001,
+      "step": 4234
+    },
+    {
+      "epoch": 34.71,
+      "learning_rate": 3e-05,
+      "loss": 0.1113,
+      "step": 4235
+    },
+    {
+      "epoch": 34.72,
+      "learning_rate": 3e-05,
+      "loss": 0.311,
+      "step": 4236
+    },
+    {
+      "epoch": 34.73,
+      "learning_rate": 3e-05,
+      "loss": 0.1593,
+      "step": 4237
+    },
+    {
+      "epoch": 34.74,
+      "learning_rate": 3e-05,
+      "loss": 0.1606,
+      "step": 4238
+    },
+    {
+      "epoch": 34.75,
+      "learning_rate": 3e-05,
+      "loss": 0.1459,
+      "step": 4239
+    },
+    {
+      "epoch": 34.75,
+      "learning_rate": 3e-05,
+      "loss": 0.1216,
+      "step": 4240
+    },
+    {
+      "epoch": 34.76,
+      "learning_rate": 3e-05,
+      "loss": 0.1823,
+      "step": 4241
+    },
+    {
+      "epoch": 34.77,
+      "learning_rate": 3e-05,
+      "loss": 0.1603,
+      "step": 4242
+    },
+    {
+      "epoch": 34.78,
+      "learning_rate": 3e-05,
+      "loss": 0.201,
+      "step": 4243
+    },
+    {
+      "epoch": 34.79,
+      "learning_rate": 3e-05,
+      "loss": 0.2003,
+      "step": 4244
+    },
+    {
+      "epoch": 34.8,
+      "learning_rate": 3e-05,
+      "loss": 0.181,
+      "step": 4245
+    },
+    {
+      "epoch": 34.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1422,
+      "step": 4246
+    },
+    {
+      "epoch": 34.81,
+      "learning_rate": 3e-05,
+      "loss": 0.1862,
+      "step": 4247
+    },
+    {
+      "epoch": 34.82,
+      "learning_rate": 3e-05,
+      "loss": 0.1713,
+      "step": 4248
+    },
+    {
+      "epoch": 34.83,
+      "learning_rate": 3e-05,
+      "loss": 0.175,
+      "step": 4249
+    },
+    {
+      "epoch": 34.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1433,
+      "step": 4250
+    },
+    {
+      "epoch": 34.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1324,
+      "step": 4251
+    },
+    {
+      "epoch": 34.85,
+      "learning_rate": 3e-05,
+      "loss": 0.2021,
+      "step": 4252
+    },
+    {
+      "epoch": 34.86,
+      "learning_rate": 3e-05,
+      "loss": 0.2222,
+      "step": 4253
+    },
+    {
+      "epoch": 34.87,
+      "learning_rate": 3e-05,
+      "loss": 0.1636,
+      "step": 4254
+    },
+    {
+      "epoch": 34.88,
+      "learning_rate": 3e-05,
+      "loss": 0.0994,
+      "step": 4255
+    },
+    {
+      "epoch": 34.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1882,
+      "step": 4256
+    },
+    {
+      "epoch": 34.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1702,
+      "step": 4257
+    },
+    {
+      "epoch": 34.9,
+      "learning_rate": 3e-05,
+      "loss": 0.1206,
+      "step": 4258
+    },
+    {
+      "epoch": 34.91,
+      "learning_rate": 3e-05,
+      "loss": 0.1099,
+      "step": 4259
+    },
+    {
+      "epoch": 34.92,
+      "learning_rate": 3e-05,
+      "loss": 0.2505,
+      "step": 4260
+    },
+    {
+      "epoch": 34.93,
+      "learning_rate": 3e-05,
+      "loss": 0.0891,
+      "step": 4261
+    },
+    {
+      "epoch": 34.93,
+      "learning_rate": 3e-05,
+      "loss": 0.3135,
+      "step": 4262
+    },
+    {
+      "epoch": 34.94,
+      "learning_rate": 3e-05,
+      "loss": 0.136,
+      "step": 4263
+    },
+    {
+      "epoch": 34.95,
+      "learning_rate": 3e-05,
+      "loss": 0.105,
+      "step": 4264
+    },
+    {
+      "epoch": 34.96,
+      "learning_rate": 3e-05,
+      "loss": 0.2001,
+      "step": 4265
+    },
+    {
+      "epoch": 34.97,
+      "learning_rate": 3e-05,
+      "loss": 0.1323,
+      "step": 4266
+    },
+    {
+      "epoch": 34.98,
+      "learning_rate": 3e-05,
+      "loss": 0.2405,
+      "step": 4267
+    },
+    {
+      "epoch": 34.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1326,
+      "step": 4268
+    },
+    {
+      "epoch": 34.99,
+      "learning_rate": 3e-05,
+      "loss": 0.1732,
+      "step": 4269
+    },
+    {
+      "epoch": 35.0,
+      "learning_rate": 3e-05,
+      "loss": 0.0903,
+      "step": 4270
+    },
+    {
+      "epoch": 35.0,
+      "eval_loss": 0.26123046875,
+      "eval_mae": 0.371337890625,
+      "eval_mse": 0.26123046875,
+      "eval_runtime": 51.0958,
+      "eval_samples_per_second": 3236.92,
+      "eval_steps_per_second": 0.274,
+      "step": 4270
+    },
+    {
+      "epoch": 35.01,
+      "learning_rate": 3e-05,
+      "loss": 0.1375,
+      "step": 4271
+    },
+    {
+      "epoch": 35.02,
+      "learning_rate": 3e-05,
+      "loss": 0.2079,
+      "step": 4272
+    },
+    {
+      "epoch": 35.02,
+      "learning_rate": 3e-05,
+      "loss": 0.2094,
+      "step": 4273
+    },
+    {
+      "epoch": 35.03,
+      "learning_rate": 3e-05,
+      "loss": 0.184,
+      "step": 4274
+    },
+    {
+      "epoch": 35.04,
+      "learning_rate": 3e-05,
+      "loss": 0.2028,
+      "step": 4275
+    },
+    {
+      "epoch": 35.05,
+      "learning_rate": 3e-05,
+      "loss": 0.1586,
+      "step": 4276
+    },
+    {
+      "epoch": 35.06,
+      "learning_rate": 3e-05,
+      "loss": 0.1615,
+      "step": 4277
+    },
+    {
+      "epoch": 35.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1711,
+      "step": 4278
+    },
+    {
+      "epoch": 35.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1003,
+      "step": 4279
+    },
+    {
+      "epoch": 35.08,
+      "learning_rate": 3e-05,
+      "loss": 0.1736,
+      "step": 4280
+    },
+    {
+      "epoch": 35.09,
+      "learning_rate": 3e-05,
+      "loss": 0.1842,
+      "step": 4281
+    },
+    {
+      "epoch": 35.1,
+      "learning_rate": 3e-05,
+      "loss": 0.2134,
+      "step": 4282
+    },
+    {
+      "epoch": 35.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1565,
+      "step": 4283
+    },
+    {
+      "epoch": 35.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1531,
+      "step": 4284
+    },
+    {
+      "epoch": 35.12,
+      "learning_rate": 3e-05,
+      "loss": 0.1154,
+      "step": 4285
+    },
+    {
+      "epoch": 35.13,
+      "learning_rate": 3e-05,
+      "loss": 0.1345,
+      "step": 4286
+    },
+    {
+      "epoch": 35.14,
+      "learning_rate": 3e-05,
+      "loss": 0.1554,
+      "step": 4287
+    },
+    {
+      "epoch": 35.15,
+      "learning_rate": 3e-05,
+      "loss": 0.1493,
+      "step": 4288
+    },
+    {
+      "epoch": 35.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1365,
+      "step": 4289
+    },
+    {
+      "epoch": 35.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1293,
+      "step": 4290
+    },
+    {
+      "epoch": 35.17,
+      "learning_rate": 3e-05,
+      "loss": 0.1545,
+      "step": 4291
+    },
+    {
+      "epoch": 35.18,
+      "learning_rate": 3e-05,
+      "loss": 0.188,
+      "step": 4292
+    },
+    {
+      "epoch": 35.19,
+      "learning_rate": 3e-05,
+      "loss": 0.2175,
+      "step": 4293
+    },
+    {
+      "epoch": 35.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1321,
+      "step": 4294
+    },
+    {
+      "epoch": 35.2,
+      "learning_rate": 3e-05,
+      "loss": 0.2445,
+      "step": 4295
+    },
+    {
+      "epoch": 35.21,
+      "learning_rate": 3e-05,
+      "loss": 0.1248,
+      "step": 4296
+    },
+    {
+      "epoch": 35.22,
+      "learning_rate": 3e-05,
+      "loss": 0.1865,
+      "step": 4297
+    },
+    {
+      "epoch": 35.23,
+      "learning_rate": 3e-05,
+      "loss": 0.1661,
+      "step": 4298
+    },
+    {
+      "epoch": 35.24,
+      "learning_rate": 3e-05,
+      "loss": 0.1039,
+      "step": 4299
+    },
+    {
+      "epoch": 35.25,
+      "learning_rate": 3e-05,
+      "loss": 0.168,
+      "step": 4300
+    },
+    {
+      "epoch": 35.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1132,
+      "step": 4301
+    },
+    {
+      "epoch": 35.26,
+      "learning_rate": 3e-05,
+      "loss": 0.1823,
+      "step": 4302
+    },
+    {
+      "epoch": 35.27,
+      "learning_rate": 3e-05,
+      "loss": 0.1803,
+      "step": 4303
+    },
+    {
+      "epoch": 35.28,
+      "learning_rate": 3e-05,
+      "loss": 0.1857,
+      "step": 4304
+    },
+    {
+      "epoch": 35.29,
+      "learning_rate": 3e-05,
+      "loss": 0.2157,
+      "step": 4305
+    },
+    {
+      "epoch": 35.3,
+      "learning_rate": 3e-05,
+      "loss": 0.2203,
+      "step": 4306
+    },
+    {
+      "epoch": 35.3,
+      "learning_rate": 3e-05,
+      "loss": 0.0876,
+      "step": 4307
+    },
+    {
+      "epoch": 35.31,
+      "learning_rate": 3e-05,
+      "loss": 0.1599,
+      "step": 4308
+    },
+    {
+      "epoch": 35.32,
+      "learning_rate": 3e-05,
+      "loss": 0.1376,
+      "step": 4309
+    },
+    {
+      "epoch": 35.33,
+      "learning_rate": 3e-05,
+      "loss": 0.2023,
+      "step": 4310
+    },
+    {
+      "epoch": 35.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1805,
+      "step": 4311
+    },
+    {
+      "epoch": 35.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1092,
+      "step": 4312
+    },
+    {
+      "epoch": 35.35,
+      "learning_rate": 3e-05,
+      "loss": 0.1265,
+      "step": 4313
+    },
+    {
+      "epoch": 35.36,
+      "learning_rate": 3e-05,
+      "loss": 0.1971,
+      "step": 4314
+    },
+    {
+      "epoch": 35.37,
+      "learning_rate": 3e-05,
+      "loss": 0.194,
+      "step": 4315
+    },
+    {
+      "epoch": 35.38,
+      "learning_rate": 3e-05,
+      "loss": 0.131,
+      "step": 4316
+    },
+    {
+      "epoch": 35.39,
+      "learning_rate": 3e-05,
+      "loss": 0.147,
+      "step": 4317
+    },
+    {
+      "epoch": 35.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1151,
+      "step": 4318
+    },
+    {
+      "epoch": 35.4,
+      "learning_rate": 3e-05,
+      "loss": 0.1956,
+      "step": 4319
+    },
+    {
+      "epoch": 35.41,
+      "learning_rate": 3e-05,
+      "loss": 0.1272,
+      "step": 4320
+    },
+    {
+      "epoch": 35.42,
+      "learning_rate": 3e-05,
+      "loss": 0.1384,
+      "step": 4321
+    },
+    {
+      "epoch": 35.43,
+      "learning_rate": 3e-05,
+      "loss": 0.135,
+      "step": 4322
+    },
+    {
+      "epoch": 35.43,
+      "learning_rate": 3e-05,
+      "loss": 0.1703,
+      "step": 4323
+    },
+    {
+      "epoch": 35.44,
+      "learning_rate": 3e-05,
+      "loss": 0.1533,
+      "step": 4324
+    },
+    {
+      "epoch": 35.45,
+      "learning_rate": 3e-05,
+      "loss": 0.2542,
+      "step": 4325
+    },
+    {
+      "epoch": 35.46,
+      "learning_rate": 3e-05,
+      "loss": 0.1484,
+      "step": 4326
+    },
+    {
+      "epoch": 35.47,
+      "learning_rate": 3e-05,
+      "loss": 0.1982,
+      "step": 4327
+    },
+    {
+      "epoch": 35.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1006,
+      "step": 4328
+    },
+    {
+      "epoch": 35.48,
+      "learning_rate": 3e-05,
+      "loss": 0.2247,
+      "step": 4329
+    },
+    {
+      "epoch": 35.49,
+      "learning_rate": 3e-05,
+      "loss": 0.099,
+      "step": 4330
+    },
+    {
+      "epoch": 35.5,
+      "learning_rate": 3e-05,
+      "loss": 0.1091,
+      "step": 4331
+    },
+    {
+      "epoch": 35.51,
+      "learning_rate": 3e-05,
+      "loss": 0.1013,
+      "step": 4332
+    },
+    {
+      "epoch": 35.52,
+      "learning_rate": 3e-05,
+      "loss": 0.146,
+      "step": 4333
+    },
+    {
+      "epoch": 35.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1989,
+      "step": 4334
+    },
+    {
+      "epoch": 35.53,
+      "learning_rate": 3e-05,
+      "loss": 0.1238,
+      "step": 4335
+    },
+    {
+      "epoch": 35.54,
+      "learning_rate": 3e-05,
+      "loss": 0.1526,
+      "step": 4336
+    },
+    {
+      "epoch": 35.55,
+      "learning_rate": 3e-05,
+      "loss": 0.1311,
+      "step": 4337
+    },
+    {
+      "epoch": 35.56,
+      "learning_rate": 3e-05,
+      "loss": 0.1255,
+      "step": 4338
+    },
+    {
+      "epoch": 35.57,
+      "learning_rate": 3e-05,
+      "loss": 0.181,
+      "step": 4339
+    },
+    {
+      "epoch": 35.57,
+      "learning_rate": 3e-05,
+      "loss": 0.1493,
+      "step": 4340
+    },
+    {
+      "epoch": 35.58,
+      "learning_rate": 3e-05,
+      "loss": 0.1031,
+      "step": 4341
+    },
+    {
+      "epoch": 35.59,
+      "learning_rate": 3e-05,
+      "loss": 0.1627,
+      "step": 4342
+    },
+    {
+      "epoch": 35.6,
+      "learning_rate": 3e-05,
+      "loss": 0.1714,
+      "step": 4343
+    },
+    {
+      "epoch": 35.61,
+      "learning_rate": 3e-05,
+      "loss": 0.136,
+      "step": 4344
+    },
+    {
+      "epoch": 35.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1149,
+      "step": 4345
+    },
+    {
+      "epoch": 35.62,
+      "learning_rate": 3e-05,
+      "loss": 0.1547,
+      "step": 4346
+    },
+    {
+      "epoch": 35.63,
+      "learning_rate": 3e-05,
+      "loss": 0.1888,
+      "step": 4347
+    },
+    {
+      "epoch": 35.64,
+      "learning_rate": 3e-05,
+      "loss": 0.1898,
+      "step": 4348
+    },
+    {
+      "epoch": 35.65,
+      "learning_rate": 3e-05,
+      "loss": 0.1198,
+      "step": 4349
+    },
+    {
+      "epoch": 35.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1317,
+      "step": 4350
+    },
+    {
+      "epoch": 35.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1116,
+      "step": 4351
+    },
+    {
+      "epoch": 35.67,
+      "learning_rate": 3e-05,
+      "loss": 0.2061,
+      "step": 4352
+    },
+    {
+      "epoch": 35.68,
+      "learning_rate": 3e-05,
+      "loss": 0.1594,
+      "step": 4353
+    },
+    {
+      "epoch": 35.69,
+      "learning_rate": 3e-05,
+      "loss": 0.1998,
+      "step": 4354
+    },
+    {
+      "epoch": 35.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1222,
+      "step": 4355
+    },
+    {
+      "epoch": 35.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1367,
+      "step": 4356
+    },
+    {
+      "epoch": 35.71,
+      "learning_rate": 3e-05,
+      "loss": 0.0829,
+      "step": 4357
+    },
+    {
+      "epoch": 35.72,
+      "learning_rate": 3e-05,
+      "loss": 0.1113,
+      "step": 4358
+    },
+    {
+      "epoch": 35.73,
+      "learning_rate": 3e-05,
+      "loss": 0.1405,
+      "step": 4359
+    },
+    {
+      "epoch": 35.74,
+      "learning_rate": 3e-05,
+      "loss": 0.1774,
+      "step": 4360
+    },
+    {
+      "epoch": 35.75,
+      "learning_rate": 3e-05,
+      "loss": 0.1632,
+      "step": 4361
+    },
+    {
+      "epoch": 35.75,
+      "learning_rate": 3e-05,
+      "loss": 0.1521,
+      "step": 4362
+    },
+    {
+      "epoch": 35.76,
+      "learning_rate": 3e-05,
+      "loss": 0.1403,
+      "step": 4363
+    },
+    {
+      "epoch": 35.77,
+      "learning_rate": 3e-05,
+      "loss": 0.1602,
+      "step": 4364
+    },
+    {
+      "epoch": 35.78,
+      "learning_rate": 3e-05,
+      "loss": 0.3291,
+      "step": 4365
+    },
+    {
+      "epoch": 35.79,
+      "learning_rate": 3e-05,
+      "loss": 0.2009,
+      "step": 4366
+    },
+    {
+      "epoch": 35.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1381,
+      "step": 4367
+    },
+    {
+      "epoch": 35.8,
+      "learning_rate": 3e-05,
+      "loss": 0.3845,
+      "step": 4368
+    },
+    {
+      "epoch": 35.81,
+      "learning_rate": 3e-05,
+      "loss": 0.2362,
+      "step": 4369
+    },
+    {
+      "epoch": 35.82,
+      "learning_rate": 3e-05,
+      "loss": 0.1152,
+      "step": 4370
+    },
+    {
+      "epoch": 35.83,
+      "learning_rate": 3e-05,
+      "loss": 0.1703,
+      "step": 4371
+    },
+    {
+      "epoch": 35.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1625,
+      "step": 4372
+    },
+    {
+      "epoch": 35.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1864,
+      "step": 4373
+    },
+    {
+      "epoch": 35.85,
+      "learning_rate": 3e-05,
+      "loss": 0.1683,
+      "step": 4374
+    },
+    {
+      "epoch": 35.86,
+      "learning_rate": 3e-05,
+      "loss": 0.1678,
+      "step": 4375
+    },
+    {
+      "epoch": 35.87,
+      "learning_rate": 3e-05,
+      "loss": 0.2177,
+      "step": 4376
+    },
+    {
+      "epoch": 35.88,
+      "learning_rate": 3e-05,
+      "loss": 0.1327,
+      "step": 4377
+    },
+    {
+      "epoch": 35.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1376,
+      "step": 4378
+    },
+    {
+      "epoch": 35.89,
+      "learning_rate": 3e-05,
+      "loss": 0.2529,
+      "step": 4379
+    },
+    {
+      "epoch": 35.9,
+      "learning_rate": 3e-05,
+      "loss": 0.178,
+      "step": 4380
+    },
+    {
+      "epoch": 35.91,
+      "learning_rate": 3e-05,
+      "loss": 0.1378,
+      "step": 4381
+    },
+    {
+      "epoch": 35.92,
+      "learning_rate": 3e-05,
+      "loss": 0.1453,
+      "step": 4382
+    },
+    {
+      "epoch": 35.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1443,
+      "step": 4383
+    },
+    {
+      "epoch": 35.93,
+      "learning_rate": 3e-05,
+      "loss": 0.167,
+      "step": 4384
+    },
+    {
+      "epoch": 35.94,
+      "learning_rate": 3e-05,
+      "loss": 0.1635,
+      "step": 4385
+    },
+    {
+      "epoch": 35.95,
+      "learning_rate": 3e-05,
+      "loss": 0.1381,
+      "step": 4386
+    },
+    {
+      "epoch": 35.96,
+      "learning_rate": 3e-05,
+      "loss": 0.135,
+      "step": 4387
+    },
+    {
+      "epoch": 35.97,
+      "learning_rate": 3e-05,
+      "loss": 0.1901,
+      "step": 4388
+    },
+    {
+      "epoch": 35.98,
+      "learning_rate": 3e-05,
+      "loss": 0.3274,
+      "step": 4389
+    },
+    {
+      "epoch": 35.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1032,
+      "step": 4390
+    },
+    {
+      "epoch": 35.99,
+      "learning_rate": 3e-05,
+      "loss": 0.196,
+      "step": 4391
+    },
+    {
+      "epoch": 36.0,
+      "learning_rate": 3e-05,
+      "loss": 0.0623,
+      "step": 4392
+    },
+    {
+      "epoch": 36.0,
+      "eval_loss": 0.261474609375,
+      "eval_mae": 0.372314453125,
+      "eval_mse": 0.261474609375,
+      "eval_runtime": 51.1314,
+      "eval_samples_per_second": 3234.663,
+      "eval_steps_per_second": 0.274,
+      "step": 4392
+    },
+    {
+      "epoch": 36.01,
+      "learning_rate": 3e-05,
+      "loss": 0.1687,
+      "step": 4393
+    },
+    {
+      "epoch": 36.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1488,
+      "step": 4394
+    },
+    {
+      "epoch": 36.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1609,
+      "step": 4395
+    },
+    {
+      "epoch": 36.03,
+      "learning_rate": 3e-05,
+      "loss": 0.147,
+      "step": 4396
+    },
+    {
+      "epoch": 36.04,
+      "learning_rate": 3e-05,
+      "loss": 0.2062,
+      "step": 4397
+    },
+    {
+      "epoch": 36.05,
+      "learning_rate": 3e-05,
+      "loss": 0.1189,
+      "step": 4398
+    },
+    {
+      "epoch": 36.06,
+      "learning_rate": 3e-05,
+      "loss": 0.2161,
+      "step": 4399
+    },
+    {
+      "epoch": 36.07,
+      "learning_rate": 3e-05,
+      "loss": 0.17,
+      "step": 4400
+    },
+    {
+      "epoch": 36.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1039,
+      "step": 4401
+    },
+    {
+      "epoch": 36.08,
+      "learning_rate": 3e-05,
+      "loss": 0.1848,
+      "step": 4402
+    },
+    {
+      "epoch": 36.09,
+      "learning_rate": 3e-05,
+      "loss": 0.1815,
+      "step": 4403
+    },
+    {
+      "epoch": 36.1,
+      "learning_rate": 3e-05,
+      "loss": 0.1228,
+      "step": 4404
+    },
+    {
+      "epoch": 36.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1592,
+      "step": 4405
+    },
+    {
+      "epoch": 36.11,
+      "learning_rate": 3e-05,
+      "loss": 0.2656,
+      "step": 4406
+    },
+    {
+      "epoch": 36.12,
+      "learning_rate": 3e-05,
+      "loss": 0.1642,
+      "step": 4407
+    },
+    {
+      "epoch": 36.13,
+      "learning_rate": 3e-05,
+      "loss": 0.1467,
+      "step": 4408
+    },
+    {
+      "epoch": 36.14,
+      "learning_rate": 3e-05,
+      "loss": 0.1908,
+      "step": 4409
+    },
+    {
+      "epoch": 36.15,
+      "learning_rate": 3e-05,
+      "loss": 0.1036,
+      "step": 4410
+    },
+    {
+      "epoch": 36.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1235,
+      "step": 4411
+    },
+    {
+      "epoch": 36.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1006,
+      "step": 4412
+    },
+    {
+      "epoch": 36.17,
+      "learning_rate": 3e-05,
+      "loss": 0.1072,
+      "step": 4413
+    },
+    {
+      "epoch": 36.18,
+      "learning_rate": 3e-05,
+      "loss": 0.1669,
+      "step": 4414
+    },
+    {
+      "epoch": 36.19,
+      "learning_rate": 3e-05,
+      "loss": 0.1519,
+      "step": 4415
+    },
+    {
+      "epoch": 36.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1708,
+      "step": 4416
+    },
+    {
+      "epoch": 36.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1768,
+      "step": 4417
+    },
+    {
+      "epoch": 36.21,
+      "learning_rate": 3e-05,
+      "loss": 0.0919,
+      "step": 4418
+    },
+    {
+      "epoch": 36.22,
+      "learning_rate": 3e-05,
+      "loss": 0.1761,
+      "step": 4419
+    },
+    {
+      "epoch": 36.23,
+      "learning_rate": 3e-05,
+      "loss": 0.3076,
+      "step": 4420
+    },
+    {
+      "epoch": 36.24,
+      "learning_rate": 3e-05,
+      "loss": 0.1376,
+      "step": 4421
+    },
+    {
+      "epoch": 36.25,
+      "learning_rate": 3e-05,
+      "loss": 0.149,
+      "step": 4422
+    },
+    {
+      "epoch": 36.25,
+      "learning_rate": 3e-05,
+      "loss": 0.2144,
+      "step": 4423
+    },
+    {
+      "epoch": 36.26,
+      "learning_rate": 3e-05,
+      "loss": 0.101,
+      "step": 4424
+    },
+    {
+      "epoch": 36.27,
+      "learning_rate": 3e-05,
+      "loss": 0.2269,
+      "step": 4425
+    },
+    {
+      "epoch": 36.28,
+      "learning_rate": 3e-05,
+      "loss": 0.2112,
+      "step": 4426
+    },
+    {
+      "epoch": 36.29,
+      "learning_rate": 3e-05,
+      "loss": 0.1365,
+      "step": 4427
+    },
+    {
+      "epoch": 36.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1686,
+      "step": 4428
+    },
+    {
+      "epoch": 36.3,
+      "learning_rate": 3e-05,
+      "loss": 0.099,
+      "step": 4429
+    },
+    {
+      "epoch": 36.31,
+      "learning_rate": 3e-05,
+      "loss": 0.1393,
+      "step": 4430
+    },
+    {
+      "epoch": 36.32,
+      "learning_rate": 3e-05,
+      "loss": 0.1338,
+      "step": 4431
+    },
+    {
+      "epoch": 36.33,
+      "learning_rate": 3e-05,
+      "loss": 0.1525,
+      "step": 4432
+    },
+    {
+      "epoch": 36.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1724,
+      "step": 4433
+    },
+    {
+      "epoch": 36.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1362,
+      "step": 4434
+    },
+    {
+      "epoch": 36.35,
+      "learning_rate": 3e-05,
+      "loss": 0.1571,
+      "step": 4435
+    },
+    {
+      "epoch": 36.36,
+      "learning_rate": 3e-05,
+      "loss": 0.213,
+      "step": 4436
+    },
+    {
+      "epoch": 36.37,
+      "learning_rate": 3e-05,
+      "loss": 0.1578,
+      "step": 4437
+    },
+    {
+      "epoch": 36.38,
+      "learning_rate": 3e-05,
+      "loss": 0.1342,
+      "step": 4438
+    },
+    {
+      "epoch": 36.39,
+      "learning_rate": 3e-05,
+      "loss": 0.2128,
+      "step": 4439
+    },
+    {
+      "epoch": 36.39,
+      "learning_rate": 3e-05,
+      "loss": 0.2275,
+      "step": 4440
+    },
+    {
+      "epoch": 36.4,
+      "learning_rate": 3e-05,
+      "loss": 0.158,
+      "step": 4441
+    },
+    {
+      "epoch": 36.41,
+      "learning_rate": 3e-05,
+      "loss": 0.1281,
+      "step": 4442
+    },
+    {
+      "epoch": 36.42,
+      "learning_rate": 3e-05,
+      "loss": 0.1708,
+      "step": 4443
+    },
+    {
+      "epoch": 36.43,
+      "learning_rate": 3e-05,
+      "loss": 0.1122,
+      "step": 4444
+    },
+    {
+      "epoch": 36.43,
+      "learning_rate": 3e-05,
+      "loss": 0.1855,
+      "step": 4445
+    },
+    {
+      "epoch": 36.44,
+      "learning_rate": 3e-05,
+      "loss": 0.1199,
+      "step": 4446
+    },
+    {
+      "epoch": 36.45,
+      "learning_rate": 3e-05,
+      "loss": 0.1608,
+      "step": 4447
+    },
+    {
+      "epoch": 36.46,
+      "learning_rate": 3e-05,
+      "loss": 0.123,
+      "step": 4448
+    },
+    {
+      "epoch": 36.47,
+      "learning_rate": 3e-05,
+      "loss": 0.1521,
+      "step": 4449
+    },
+    {
+      "epoch": 36.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1007,
+      "step": 4450
+    },
+    {
+      "epoch": 36.48,
+      "learning_rate": 3e-05,
+      "loss": 0.2198,
+      "step": 4451
+    },
+    {
+      "epoch": 36.49,
+      "learning_rate": 3e-05,
+      "loss": 0.0936,
+      "step": 4452
+    },
+    {
+      "epoch": 36.5,
+      "learning_rate": 3e-05,
+      "loss": 0.1675,
+      "step": 4453
+    },
+    {
+      "epoch": 36.51,
+      "learning_rate": 3e-05,
+      "loss": 0.1455,
+      "step": 4454
+    },
+    {
+      "epoch": 36.52,
+      "learning_rate": 3e-05,
+      "loss": 0.0823,
+      "step": 4455
+    },
+    {
+      "epoch": 36.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1888,
+      "step": 4456
+    },
+    {
+      "epoch": 36.53,
+      "learning_rate": 3e-05,
+      "loss": 0.1948,
+      "step": 4457
+    },
+    {
+      "epoch": 36.54,
+      "learning_rate": 3e-05,
+      "loss": 0.1211,
+      "step": 4458
+    },
+    {
+      "epoch": 36.55,
+      "learning_rate": 3e-05,
+      "loss": 0.0926,
+      "step": 4459
+    },
+    {
+      "epoch": 36.56,
+      "learning_rate": 3e-05,
+      "loss": 0.1345,
+      "step": 4460
+    },
+    {
+      "epoch": 36.57,
+      "learning_rate": 3e-05,
+      "loss": 0.1649,
+      "step": 4461
+    },
+    {
+      "epoch": 36.57,
+      "learning_rate": 3e-05,
+      "loss": 0.11,
+      "step": 4462
+    },
+    {
+      "epoch": 36.58,
+      "learning_rate": 3e-05,
+      "loss": 0.1255,
+      "step": 4463
+    },
+    {
+      "epoch": 36.59,
+      "learning_rate": 3e-05,
+      "loss": 0.1238,
+      "step": 4464
+    },
+    {
+      "epoch": 36.6,
+      "learning_rate": 3e-05,
+      "loss": 0.207,
+      "step": 4465
+    },
+    {
+      "epoch": 36.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1451,
+      "step": 4466
+    },
+    {
+      "epoch": 36.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1688,
+      "step": 4467
+    },
+    {
+      "epoch": 36.62,
+      "learning_rate": 3e-05,
+      "loss": 0.123,
+      "step": 4468
+    },
+    {
+      "epoch": 36.63,
+      "learning_rate": 3e-05,
+      "loss": 0.1504,
+      "step": 4469
+    },
+    {
+      "epoch": 36.64,
+      "learning_rate": 3e-05,
+      "loss": 0.1087,
+      "step": 4470
+    },
+    {
+      "epoch": 36.65,
+      "learning_rate": 3e-05,
+      "loss": 0.1154,
+      "step": 4471
+    },
+    {
+      "epoch": 36.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1917,
+      "step": 4472
+    },
+    {
+      "epoch": 36.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1047,
+      "step": 4473
+    },
+    {
+      "epoch": 36.67,
+      "learning_rate": 3e-05,
+      "loss": 0.293,
+      "step": 4474
+    },
+    {
+      "epoch": 36.68,
+      "learning_rate": 3e-05,
+      "loss": 0.2104,
+      "step": 4475
+    },
+    {
+      "epoch": 36.69,
+      "learning_rate": 3e-05,
+      "loss": 0.1633,
+      "step": 4476
+    },
+    {
+      "epoch": 36.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1281,
+      "step": 4477
+    },
+    {
+      "epoch": 36.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1006,
+      "step": 4478
+    },
+    {
+      "epoch": 36.71,
+      "learning_rate": 3e-05,
+      "loss": 0.1238,
+      "step": 4479
+    },
+    {
+      "epoch": 36.72,
+      "learning_rate": 3e-05,
+      "loss": 0.1323,
+      "step": 4480
+    },
+    {
+      "epoch": 36.73,
+      "learning_rate": 3e-05,
+      "loss": 0.161,
+      "step": 4481
+    },
+    {
+      "epoch": 36.74,
+      "learning_rate": 3e-05,
+      "loss": 0.2162,
+      "step": 4482
+    },
+    {
+      "epoch": 36.75,
+      "learning_rate": 3e-05,
+      "loss": 0.158,
+      "step": 4483
+    },
+    {
+      "epoch": 36.75,
+      "learning_rate": 3e-05,
+      "loss": 0.1437,
+      "step": 4484
+    },
+    {
+      "epoch": 36.76,
+      "learning_rate": 3e-05,
+      "loss": 0.1478,
+      "step": 4485
+    },
+    {
+      "epoch": 36.77,
+      "learning_rate": 3e-05,
+      "loss": 0.2137,
+      "step": 4486
+    },
+    {
+      "epoch": 36.78,
+      "learning_rate": 3e-05,
+      "loss": 0.1405,
+      "step": 4487
+    },
+    {
+      "epoch": 36.79,
+      "learning_rate": 3e-05,
+      "loss": 0.1514,
+      "step": 4488
+    },
+    {
+      "epoch": 36.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1699,
+      "step": 4489
+    },
+    {
+      "epoch": 36.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1625,
+      "step": 4490
+    },
+    {
+      "epoch": 36.81,
+      "learning_rate": 3e-05,
+      "loss": 0.1014,
+      "step": 4491
+    },
+    {
+      "epoch": 36.82,
+      "learning_rate": 3e-05,
+      "loss": 0.1383,
+      "step": 4492
+    },
+    {
+      "epoch": 36.83,
+      "learning_rate": 3e-05,
+      "loss": 0.1603,
+      "step": 4493
+    },
+    {
+      "epoch": 36.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1847,
+      "step": 4494
+    },
+    {
+      "epoch": 36.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1359,
+      "step": 4495
+    },
+    {
+      "epoch": 36.85,
+      "learning_rate": 3e-05,
+      "loss": 0.1191,
+      "step": 4496
+    },
+    {
+      "epoch": 36.86,
+      "learning_rate": 3e-05,
+      "loss": 0.0951,
+      "step": 4497
+    },
+    {
+      "epoch": 36.87,
+      "learning_rate": 3e-05,
+      "loss": 0.1183,
+      "step": 4498
+    },
+    {
+      "epoch": 36.88,
+      "learning_rate": 3e-05,
+      "loss": 0.1147,
+      "step": 4499
+    },
+    {
+      "epoch": 36.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1556,
+      "step": 4500
+    },
+    {
+      "epoch": 36.89,
+      "learning_rate": 3e-05,
+      "loss": 0.2225,
+      "step": 4501
+    },
+    {
+      "epoch": 36.9,
+      "learning_rate": 3e-05,
+      "loss": 0.1378,
+      "step": 4502
+    },
+    {
+      "epoch": 36.91,
+      "learning_rate": 3e-05,
+      "loss": 0.1166,
+      "step": 4503
+    },
+    {
+      "epoch": 36.92,
+      "learning_rate": 3e-05,
+      "loss": 0.1296,
+      "step": 4504
+    },
+    {
+      "epoch": 36.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1821,
+      "step": 4505
+    },
+    {
+      "epoch": 36.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1182,
+      "step": 4506
+    },
+    {
+      "epoch": 36.94,
+      "learning_rate": 3e-05,
+      "loss": 0.1769,
+      "step": 4507
+    },
+    {
+      "epoch": 36.95,
+      "learning_rate": 3e-05,
+      "loss": 0.1104,
+      "step": 4508
+    },
+    {
+      "epoch": 36.96,
+      "learning_rate": 3e-05,
+      "loss": 0.1067,
+      "step": 4509
+    },
+    {
+      "epoch": 36.97,
+      "learning_rate": 3e-05,
+      "loss": 0.1571,
+      "step": 4510
+    },
+    {
+      "epoch": 36.98,
+      "learning_rate": 3e-05,
+      "loss": 0.158,
+      "step": 4511
+    },
+    {
+      "epoch": 36.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1389,
+      "step": 4512
+    },
+    {
+      "epoch": 36.99,
+      "learning_rate": 3e-05,
+      "loss": 0.22,
+      "step": 4513
+    },
+    {
+      "epoch": 37.0,
+      "learning_rate": 3e-05,
+      "loss": 0.1057,
+      "step": 4514
+    },
+    {
+      "epoch": 37.0,
+      "eval_loss": 0.260009765625,
+      "eval_mae": 0.369384765625,
+      "eval_mse": 0.260009765625,
+      "eval_runtime": 51.1147,
+      "eval_samples_per_second": 3235.721,
+      "eval_steps_per_second": 0.274,
+      "step": 4514
+    },
+    {
+      "epoch": 37.01,
+      "learning_rate": 3e-05,
+      "loss": 0.103,
+      "step": 4515
+    },
+    {
+      "epoch": 37.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1359,
+      "step": 4516
+    },
+    {
+      "epoch": 37.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1415,
+      "step": 4517
+    },
+    {
+      "epoch": 37.03,
+      "learning_rate": 3e-05,
+      "loss": 0.1479,
+      "step": 4518
+    },
+    {
+      "epoch": 37.04,
+      "learning_rate": 3e-05,
+      "loss": 0.1741,
+      "step": 4519
+    },
+    {
+      "epoch": 37.05,
+      "learning_rate": 3e-05,
+      "loss": 0.1376,
+      "step": 4520
+    },
+    {
+      "epoch": 37.06,
+      "learning_rate": 3e-05,
+      "loss": 0.3284,
+      "step": 4521
+    },
+    {
+      "epoch": 37.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1433,
+      "step": 4522
+    },
+    {
+      "epoch": 37.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1481,
+      "step": 4523
+    },
+    {
+      "epoch": 37.08,
+      "learning_rate": 3e-05,
+      "loss": 0.1141,
+      "step": 4524
+    },
+    {
+      "epoch": 37.09,
+      "learning_rate": 3e-05,
+      "loss": 0.163,
+      "step": 4525
+    },
+    {
+      "epoch": 37.1,
+      "learning_rate": 3e-05,
+      "loss": 0.1333,
+      "step": 4526
+    },
+    {
+      "epoch": 37.11,
+      "learning_rate": 3e-05,
+      "loss": 0.2145,
+      "step": 4527
+    },
+    {
+      "epoch": 37.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1312,
+      "step": 4528
+    },
+    {
+      "epoch": 37.12,
+      "learning_rate": 3e-05,
+      "loss": 0.1218,
+      "step": 4529
+    },
+    {
+      "epoch": 37.13,
+      "learning_rate": 3e-05,
+      "loss": 0.134,
+      "step": 4530
+    },
+    {
+      "epoch": 37.14,
+      "learning_rate": 3e-05,
+      "loss": 0.2399,
+      "step": 4531
+    },
+    {
+      "epoch": 37.15,
+      "learning_rate": 3e-05,
+      "loss": 0.1697,
+      "step": 4532
+    },
+    {
+      "epoch": 37.16,
+      "learning_rate": 3e-05,
+      "loss": 0.2502,
+      "step": 4533
+    },
+    {
+      "epoch": 37.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1108,
+      "step": 4534
+    },
+    {
+      "epoch": 37.17,
+      "learning_rate": 3e-05,
+      "loss": 0.1315,
+      "step": 4535
+    },
+    {
+      "epoch": 37.18,
+      "learning_rate": 3e-05,
+      "loss": 0.1722,
+      "step": 4536
+    },
+    {
+      "epoch": 37.19,
+      "learning_rate": 3e-05,
+      "loss": 0.1411,
+      "step": 4537
+    },
+    {
+      "epoch": 37.2,
+      "learning_rate": 3e-05,
+      "loss": 0.181,
+      "step": 4538
+    },
+    {
+      "epoch": 37.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1646,
+      "step": 4539
+    },
+    {
+      "epoch": 37.21,
+      "learning_rate": 3e-05,
+      "loss": 0.1646,
+      "step": 4540
+    },
+    {
+      "epoch": 37.22,
+      "learning_rate": 3e-05,
+      "loss": 0.1505,
+      "step": 4541
+    },
+    {
+      "epoch": 37.23,
+      "learning_rate": 3e-05,
+      "loss": 0.1913,
+      "step": 4542
+    },
+    {
+      "epoch": 37.24,
+      "learning_rate": 3e-05,
+      "loss": 0.1809,
+      "step": 4543
+    },
+    {
+      "epoch": 37.25,
+      "learning_rate": 3e-05,
+      "loss": 0.2781,
+      "step": 4544
+    },
+    {
+      "epoch": 37.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1768,
+      "step": 4545
+    },
+    {
+      "epoch": 37.26,
+      "learning_rate": 3e-05,
+      "loss": 0.1509,
+      "step": 4546
+    },
+    {
+      "epoch": 37.27,
+      "learning_rate": 3e-05,
+      "loss": 0.1134,
+      "step": 4547
+    },
+    {
+      "epoch": 37.28,
+      "learning_rate": 3e-05,
+      "loss": 0.1559,
+      "step": 4548
+    },
+    {
+      "epoch": 37.29,
+      "learning_rate": 3e-05,
+      "loss": 0.1439,
+      "step": 4549
+    },
+    {
+      "epoch": 37.3,
+      "learning_rate": 3e-05,
+      "loss": 0.0705,
+      "step": 4550
+    },
+    {
+      "epoch": 37.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1582,
+      "step": 4551
+    },
+    {
+      "epoch": 37.31,
+      "learning_rate": 3e-05,
+      "loss": 0.127,
+      "step": 4552
+    },
+    {
+      "epoch": 37.32,
+      "learning_rate": 3e-05,
+      "loss": 0.1357,
+      "step": 4553
+    },
+    {
+      "epoch": 37.33,
+      "learning_rate": 3e-05,
+      "loss": 0.2405,
+      "step": 4554
+    },
+    {
+      "epoch": 37.34,
+      "learning_rate": 3e-05,
+      "loss": 0.131,
+      "step": 4555
+    },
+    {
+      "epoch": 37.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1582,
+      "step": 4556
+    },
+    {
+      "epoch": 37.35,
+      "learning_rate": 3e-05,
+      "loss": 0.1631,
+      "step": 4557
+    },
+    {
+      "epoch": 37.36,
+      "learning_rate": 3e-05,
+      "loss": 0.2009,
+      "step": 4558
+    },
+    {
+      "epoch": 37.37,
+      "learning_rate": 3e-05,
+      "loss": 0.1321,
+      "step": 4559
+    },
+    {
+      "epoch": 37.38,
+      "learning_rate": 3e-05,
+      "loss": 0.1515,
+      "step": 4560
+    },
+    {
+      "epoch": 37.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1671,
+      "step": 4561
+    },
+    {
+      "epoch": 37.39,
+      "learning_rate": 3e-05,
+      "loss": 0.13,
+      "step": 4562
+    },
+    {
+      "epoch": 37.4,
+      "learning_rate": 3e-05,
+      "loss": 0.1499,
+      "step": 4563
+    },
+    {
+      "epoch": 37.41,
+      "learning_rate": 3e-05,
+      "loss": 0.1409,
+      "step": 4564
+    },
+    {
+      "epoch": 37.42,
+      "learning_rate": 3e-05,
+      "loss": 0.2113,
+      "step": 4565
+    },
+    {
+      "epoch": 37.43,
+      "learning_rate": 3e-05,
+      "loss": 0.1371,
+      "step": 4566
+    },
+    {
+      "epoch": 37.43,
+      "learning_rate": 3e-05,
+      "loss": 0.1309,
+      "step": 4567
+    },
+    {
+      "epoch": 37.44,
+      "learning_rate": 3e-05,
+      "loss": 0.1761,
+      "step": 4568
+    },
+    {
+      "epoch": 37.45,
+      "learning_rate": 3e-05,
+      "loss": 0.1089,
+      "step": 4569
+    },
+    {
+      "epoch": 37.46,
+      "learning_rate": 3e-05,
+      "loss": 0.1619,
+      "step": 4570
+    },
+    {
+      "epoch": 37.47,
+      "learning_rate": 3e-05,
+      "loss": 0.1283,
+      "step": 4571
+    },
+    {
+      "epoch": 37.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1865,
+      "step": 4572
+    },
+    {
+      "epoch": 37.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1226,
+      "step": 4573
+    },
+    {
+      "epoch": 37.49,
+      "learning_rate": 3e-05,
+      "loss": 0.1587,
+      "step": 4574
+    },
+    {
+      "epoch": 37.5,
+      "learning_rate": 3e-05,
+      "loss": 0.1489,
+      "step": 4575
+    },
+    {
+      "epoch": 37.51,
+      "learning_rate": 3e-05,
+      "loss": 0.1181,
+      "step": 4576
+    },
+    {
+      "epoch": 37.52,
+      "learning_rate": 3e-05,
+      "loss": 0.114,
+      "step": 4577
+    },
+    {
+      "epoch": 37.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1522,
+      "step": 4578
+    },
+    {
+      "epoch": 37.53,
+      "learning_rate": 3e-05,
+      "loss": 0.1549,
+      "step": 4579
+    },
+    {
+      "epoch": 37.54,
+      "learning_rate": 3e-05,
+      "loss": 0.1339,
+      "step": 4580
+    },
+    {
+      "epoch": 37.55,
+      "learning_rate": 3e-05,
+      "loss": 0.1519,
+      "step": 4581
+    },
+    {
+      "epoch": 37.56,
+      "learning_rate": 3e-05,
+      "loss": 0.1855,
+      "step": 4582
+    },
+    {
+      "epoch": 37.57,
+      "learning_rate": 3e-05,
+      "loss": 0.2413,
+      "step": 4583
+    },
+    {
+      "epoch": 37.57,
+      "learning_rate": 3e-05,
+      "loss": 0.3225,
+      "step": 4584
+    },
+    {
+      "epoch": 37.58,
+      "learning_rate": 3e-05,
+      "loss": 0.1119,
+      "step": 4585
+    },
+    {
+      "epoch": 37.59,
+      "learning_rate": 3e-05,
+      "loss": 0.2186,
+      "step": 4586
+    },
+    {
+      "epoch": 37.6,
+      "learning_rate": 3e-05,
+      "loss": 0.1272,
+      "step": 4587
+    },
+    {
+      "epoch": 37.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1674,
+      "step": 4588
+    },
+    {
+      "epoch": 37.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1259,
+      "step": 4589
+    },
+    {
+      "epoch": 37.62,
+      "learning_rate": 3e-05,
+      "loss": 0.1323,
+      "step": 4590
+    },
+    {
+      "epoch": 37.63,
+      "learning_rate": 3e-05,
+      "loss": 0.1437,
+      "step": 4591
+    },
+    {
+      "epoch": 37.64,
+      "learning_rate": 3e-05,
+      "loss": 0.1569,
+      "step": 4592
+    },
+    {
+      "epoch": 37.65,
+      "learning_rate": 3e-05,
+      "loss": 0.1448,
+      "step": 4593
+    },
+    {
+      "epoch": 37.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1532,
+      "step": 4594
+    },
+    {
+      "epoch": 37.66,
+      "learning_rate": 3e-05,
+      "loss": 0.219,
+      "step": 4595
+    },
+    {
+      "epoch": 37.67,
+      "learning_rate": 3e-05,
+      "loss": 0.0992,
+      "step": 4596
+    },
+    {
+      "epoch": 37.68,
+      "learning_rate": 3e-05,
+      "loss": 0.17,
+      "step": 4597
+    },
+    {
+      "epoch": 37.69,
+      "learning_rate": 3e-05,
+      "loss": 0.135,
+      "step": 4598
+    },
+    {
+      "epoch": 37.7,
+      "learning_rate": 3e-05,
+      "loss": 0.14,
+      "step": 4599
+    },
+    {
+      "epoch": 37.7,
+      "learning_rate": 3e-05,
+      "loss": 0.212,
+      "step": 4600
+    },
+    {
+      "epoch": 37.71,
+      "learning_rate": 3e-05,
+      "loss": 0.1691,
+      "step": 4601
+    },
+    {
+      "epoch": 37.72,
+      "learning_rate": 3e-05,
+      "loss": 0.1649,
+      "step": 4602
+    },
+    {
+      "epoch": 37.73,
+      "learning_rate": 3e-05,
+      "loss": 0.1361,
+      "step": 4603
+    },
+    {
+      "epoch": 37.74,
+      "learning_rate": 3e-05,
+      "loss": 0.1403,
+      "step": 4604
+    },
+    {
+      "epoch": 37.75,
+      "learning_rate": 3e-05,
+      "loss": 0.1161,
+      "step": 4605
+    },
+    {
+      "epoch": 37.75,
+      "learning_rate": 3e-05,
+      "loss": 0.1053,
+      "step": 4606
+    },
+    {
+      "epoch": 37.76,
+      "learning_rate": 3e-05,
+      "loss": 0.1021,
+      "step": 4607
+    },
+    {
+      "epoch": 37.77,
+      "learning_rate": 3e-05,
+      "loss": 0.1836,
+      "step": 4608
+    },
+    {
+      "epoch": 37.78,
+      "learning_rate": 3e-05,
+      "loss": 0.1265,
+      "step": 4609
+    },
+    {
+      "epoch": 37.79,
+      "learning_rate": 3e-05,
+      "loss": 0.1561,
+      "step": 4610
+    },
+    {
+      "epoch": 37.8,
+      "learning_rate": 3e-05,
+      "loss": 0.0914,
+      "step": 4611
+    },
+    {
+      "epoch": 37.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1991,
+      "step": 4612
+    },
+    {
+      "epoch": 37.81,
+      "learning_rate": 3e-05,
+      "loss": 0.1616,
+      "step": 4613
+    },
+    {
+      "epoch": 37.82,
+      "learning_rate": 3e-05,
+      "loss": 0.1042,
+      "step": 4614
+    },
+    {
+      "epoch": 37.83,
+      "learning_rate": 3e-05,
+      "loss": 0.1521,
+      "step": 4615
+    },
+    {
+      "epoch": 37.84,
+      "learning_rate": 3e-05,
+      "loss": 0.252,
+      "step": 4616
+    },
+    {
+      "epoch": 37.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1323,
+      "step": 4617
+    },
+    {
+      "epoch": 37.85,
+      "learning_rate": 3e-05,
+      "loss": 0.0833,
+      "step": 4618
+    },
+    {
+      "epoch": 37.86,
+      "learning_rate": 3e-05,
+      "loss": 0.1359,
+      "step": 4619
+    },
+    {
+      "epoch": 37.87,
+      "learning_rate": 3e-05,
+      "loss": 0.1422,
+      "step": 4620
+    },
+    {
+      "epoch": 37.88,
+      "learning_rate": 3e-05,
+      "loss": 0.145,
+      "step": 4621
+    },
+    {
+      "epoch": 37.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1315,
+      "step": 4622
+    },
+    {
+      "epoch": 37.89,
+      "learning_rate": 3e-05,
+      "loss": 0.2286,
+      "step": 4623
+    },
+    {
+      "epoch": 37.9,
+      "learning_rate": 3e-05,
+      "loss": 0.1735,
+      "step": 4624
+    },
+    {
+      "epoch": 37.91,
+      "learning_rate": 3e-05,
+      "loss": 0.1302,
+      "step": 4625
+    },
+    {
+      "epoch": 37.92,
+      "learning_rate": 3e-05,
+      "loss": 0.1586,
+      "step": 4626
+    },
+    {
+      "epoch": 37.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1624,
+      "step": 4627
+    },
+    {
+      "epoch": 37.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1487,
+      "step": 4628
+    },
+    {
+      "epoch": 37.94,
+      "learning_rate": 3e-05,
+      "loss": 0.1168,
+      "step": 4629
+    },
+    {
+      "epoch": 37.95,
+      "learning_rate": 3e-05,
+      "loss": 0.0929,
+      "step": 4630
+    },
+    {
+      "epoch": 37.96,
+      "learning_rate": 3e-05,
+      "loss": 0.2844,
+      "step": 4631
+    },
+    {
+      "epoch": 37.97,
+      "learning_rate": 3e-05,
+      "loss": 0.167,
+      "step": 4632
+    },
+    {
+      "epoch": 37.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1669,
+      "step": 4633
+    },
+    {
+      "epoch": 37.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1293,
+      "step": 4634
+    },
+    {
+      "epoch": 37.99,
+      "learning_rate": 3e-05,
+      "loss": 0.1367,
+      "step": 4635
+    },
+    {
+      "epoch": 38.0,
+      "learning_rate": 3e-05,
+      "loss": 0.0328,
+      "step": 4636
+    },
+    {
+      "epoch": 38.0,
+      "eval_loss": 0.261474609375,
+      "eval_mae": 0.370849609375,
+      "eval_mse": 0.261474609375,
+      "eval_runtime": 51.1779,
+      "eval_samples_per_second": 3231.726,
+      "eval_steps_per_second": 0.274,
+      "step": 4636
+    },
+    {
+      "epoch": 38.01,
+      "learning_rate": 3e-05,
+      "loss": 0.2039,
+      "step": 4637
+    },
+    {
+      "epoch": 38.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1195,
+      "step": 4638
+    },
+    {
+      "epoch": 38.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1635,
+      "step": 4639
+    },
+    {
+      "epoch": 38.03,
+      "learning_rate": 3e-05,
+      "loss": 0.1317,
+      "step": 4640
+    },
+    {
+      "epoch": 38.04,
+      "learning_rate": 3e-05,
+      "loss": 0.1415,
+      "step": 4641
+    },
+    {
+      "epoch": 38.05,
+      "learning_rate": 3e-05,
+      "loss": 0.2441,
+      "step": 4642
+    },
+    {
+      "epoch": 38.06,
+      "learning_rate": 3e-05,
+      "loss": 0.114,
+      "step": 4643
+    },
+    {
+      "epoch": 38.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1209,
+      "step": 4644
+    },
+    {
+      "epoch": 38.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1761,
+      "step": 4645
+    },
+    {
+      "epoch": 38.08,
+      "learning_rate": 3e-05,
+      "loss": 0.1561,
+      "step": 4646
+    },
+    {
+      "epoch": 38.09,
+      "learning_rate": 3e-05,
+      "loss": 0.1974,
+      "step": 4647
+    },
+    {
+      "epoch": 38.1,
+      "learning_rate": 3e-05,
+      "loss": 0.1868,
+      "step": 4648
+    },
+    {
+      "epoch": 38.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1426,
+      "step": 4649
+    },
+    {
+      "epoch": 38.11,
+      "learning_rate": 3e-05,
+      "loss": 0.0942,
+      "step": 4650
+    },
+    {
+      "epoch": 38.12,
+      "learning_rate": 3e-05,
+      "loss": 0.1958,
+      "step": 4651
+    },
+    {
+      "epoch": 38.13,
+      "learning_rate": 3e-05,
+      "loss": 0.2051,
+      "step": 4652
+    },
+    {
+      "epoch": 38.14,
+      "learning_rate": 3e-05,
+      "loss": 0.1088,
+      "step": 4653
+    },
+    {
+      "epoch": 38.15,
+      "learning_rate": 3e-05,
+      "loss": 0.1124,
+      "step": 4654
+    },
+    {
+      "epoch": 38.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1377,
+      "step": 4655
+    },
+    {
+      "epoch": 38.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1893,
+      "step": 4656
+    },
+    {
+      "epoch": 38.17,
+      "learning_rate": 3e-05,
+      "loss": 0.1425,
+      "step": 4657
+    },
+    {
+      "epoch": 38.18,
+      "learning_rate": 3e-05,
+      "loss": 0.1466,
+      "step": 4658
+    },
+    {
+      "epoch": 38.19,
+      "learning_rate": 3e-05,
+      "loss": 0.2361,
+      "step": 4659
+    },
+    {
+      "epoch": 38.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1581,
+      "step": 4660
+    },
+    {
+      "epoch": 38.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1664,
+      "step": 4661
+    },
+    {
+      "epoch": 38.21,
+      "learning_rate": 3e-05,
+      "loss": 0.0796,
+      "step": 4662
+    },
+    {
+      "epoch": 38.22,
+      "learning_rate": 3e-05,
+      "loss": 0.1401,
+      "step": 4663
+    },
+    {
+      "epoch": 38.23,
+      "learning_rate": 3e-05,
+      "loss": 0.1659,
+      "step": 4664
+    },
+    {
+      "epoch": 38.24,
+      "learning_rate": 3e-05,
+      "loss": 0.1267,
+      "step": 4665
+    },
+    {
+      "epoch": 38.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1434,
+      "step": 4666
+    },
+    {
+      "epoch": 38.25,
+      "learning_rate": 3e-05,
+      "loss": 0.2229,
+      "step": 4667
+    },
+    {
+      "epoch": 38.26,
+      "learning_rate": 3e-05,
+      "loss": 0.1635,
+      "step": 4668
+    },
+    {
+      "epoch": 38.27,
+      "learning_rate": 3e-05,
+      "loss": 0.1713,
+      "step": 4669
+    },
+    {
+      "epoch": 38.28,
+      "learning_rate": 3e-05,
+      "loss": 0.1561,
+      "step": 4670
+    },
+    {
+      "epoch": 38.29,
+      "learning_rate": 3e-05,
+      "loss": 0.123,
+      "step": 4671
+    },
+    {
+      "epoch": 38.3,
+      "learning_rate": 3e-05,
+      "loss": 0.2037,
+      "step": 4672
+    },
+    {
+      "epoch": 38.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1415,
+      "step": 4673
+    },
+    {
+      "epoch": 38.31,
+      "learning_rate": 3e-05,
+      "loss": 0.1099,
+      "step": 4674
+    },
+    {
+      "epoch": 38.32,
+      "learning_rate": 3e-05,
+      "loss": 0.1208,
+      "step": 4675
+    },
+    {
+      "epoch": 38.33,
+      "learning_rate": 3e-05,
+      "loss": 0.1285,
+      "step": 4676
+    },
+    {
+      "epoch": 38.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1722,
+      "step": 4677
+    },
+    {
+      "epoch": 38.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1185,
+      "step": 4678
+    },
+    {
+      "epoch": 38.35,
+      "learning_rate": 3e-05,
+      "loss": 0.1281,
+      "step": 4679
+    },
+    {
+      "epoch": 38.36,
+      "learning_rate": 3e-05,
+      "loss": 0.1298,
+      "step": 4680
+    },
+    {
+      "epoch": 38.37,
+      "learning_rate": 3e-05,
+      "loss": 0.2422,
+      "step": 4681
+    },
+    {
+      "epoch": 38.38,
+      "learning_rate": 3e-05,
+      "loss": 0.1506,
+      "step": 4682
+    },
+    {
+      "epoch": 38.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1482,
+      "step": 4683
+    },
+    {
+      "epoch": 38.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1925,
+      "step": 4684
+    },
+    {
+      "epoch": 38.4,
+      "learning_rate": 3e-05,
+      "loss": 0.0778,
+      "step": 4685
+    },
+    {
+      "epoch": 38.41,
+      "learning_rate": 3e-05,
+      "loss": 0.118,
+      "step": 4686
+    },
+    {
+      "epoch": 38.42,
+      "learning_rate": 3e-05,
+      "loss": 0.1656,
+      "step": 4687
+    },
+    {
+      "epoch": 38.43,
+      "learning_rate": 3e-05,
+      "loss": 0.1459,
+      "step": 4688
+    },
+    {
+      "epoch": 38.43,
+      "learning_rate": 3e-05,
+      "loss": 0.1078,
+      "step": 4689
+    },
+    {
+      "epoch": 38.44,
+      "learning_rate": 3e-05,
+      "loss": 0.1694,
+      "step": 4690
+    },
+    {
+      "epoch": 38.45,
+      "learning_rate": 3e-05,
+      "loss": 0.1263,
+      "step": 4691
+    },
+    {
+      "epoch": 38.46,
+      "learning_rate": 3e-05,
+      "loss": 0.1958,
+      "step": 4692
+    },
+    {
+      "epoch": 38.47,
+      "learning_rate": 3e-05,
+      "loss": 0.1022,
+      "step": 4693
+    },
+    {
+      "epoch": 38.48,
+      "learning_rate": 3e-05,
+      "loss": 0.2123,
+      "step": 4694
+    },
+    {
+      "epoch": 38.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1588,
+      "step": 4695
+    },
+    {
+      "epoch": 38.49,
+      "learning_rate": 3e-05,
+      "loss": 0.1403,
+      "step": 4696
+    },
+    {
+      "epoch": 38.5,
+      "learning_rate": 3e-05,
+      "loss": 0.1763,
+      "step": 4697
+    },
+    {
+      "epoch": 38.51,
+      "learning_rate": 3e-05,
+      "loss": 0.1481,
+      "step": 4698
+    },
+    {
+      "epoch": 38.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1429,
+      "step": 4699
+    },
+    {
+      "epoch": 38.52,
+      "learning_rate": 3e-05,
+      "loss": 0.126,
+      "step": 4700
+    },
+    {
+      "epoch": 38.53,
+      "learning_rate": 3e-05,
+      "loss": 0.1672,
+      "step": 4701
+    },
+    {
+      "epoch": 38.54,
+      "learning_rate": 3e-05,
+      "loss": 0.2125,
+      "step": 4702
+    },
+    {
+      "epoch": 38.55,
+      "learning_rate": 3e-05,
+      "loss": 0.1137,
+      "step": 4703
+    },
+    {
+      "epoch": 38.56,
+      "learning_rate": 3e-05,
+      "loss": 0.1322,
+      "step": 4704
+    },
+    {
+      "epoch": 38.57,
+      "learning_rate": 3e-05,
+      "loss": 0.1361,
+      "step": 4705
+    },
+    {
+      "epoch": 38.57,
+      "learning_rate": 3e-05,
+      "loss": 0.2253,
+      "step": 4706
+    },
+    {
+      "epoch": 38.58,
+      "learning_rate": 3e-05,
+      "loss": 0.1891,
+      "step": 4707
+    },
+    {
+      "epoch": 38.59,
+      "learning_rate": 3e-05,
+      "loss": 0.2007,
+      "step": 4708
+    },
+    {
+      "epoch": 38.6,
+      "learning_rate": 3e-05,
+      "loss": 0.155,
+      "step": 4709
+    },
+    {
+      "epoch": 38.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1306,
+      "step": 4710
+    },
+    {
+      "epoch": 38.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1224,
+      "step": 4711
+    },
+    {
+      "epoch": 38.62,
+      "learning_rate": 3e-05,
+      "loss": 0.1096,
+      "step": 4712
+    },
+    {
+      "epoch": 38.63,
+      "learning_rate": 3e-05,
+      "loss": 0.1636,
+      "step": 4713
+    },
+    {
+      "epoch": 38.64,
+      "learning_rate": 3e-05,
+      "loss": 0.1562,
+      "step": 4714
+    },
+    {
+      "epoch": 38.65,
+      "learning_rate": 3e-05,
+      "loss": 0.1984,
+      "step": 4715
+    },
+    {
+      "epoch": 38.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1433,
+      "step": 4716
+    },
+    {
+      "epoch": 38.66,
+      "learning_rate": 3e-05,
+      "loss": 0.0732,
+      "step": 4717
+    },
+    {
+      "epoch": 38.67,
+      "learning_rate": 3e-05,
+      "loss": 0.1448,
+      "step": 4718
+    },
+    {
+      "epoch": 38.68,
+      "learning_rate": 3e-05,
+      "loss": 0.1766,
+      "step": 4719
+    },
+    {
+      "epoch": 38.69,
+      "learning_rate": 3e-05,
+      "loss": 0.1575,
+      "step": 4720
+    },
+    {
+      "epoch": 38.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1105,
+      "step": 4721
+    },
+    {
+      "epoch": 38.7,
+      "learning_rate": 3e-05,
+      "loss": 0.2292,
+      "step": 4722
+    },
+    {
+      "epoch": 38.71,
+      "learning_rate": 3e-05,
+      "loss": 0.153,
+      "step": 4723
+    },
+    {
+      "epoch": 38.72,
+      "learning_rate": 3e-05,
+      "loss": 0.1602,
+      "step": 4724
+    },
+    {
+      "epoch": 38.73,
+      "learning_rate": 3e-05,
+      "loss": 0.1287,
+      "step": 4725
+    },
+    {
+      "epoch": 38.74,
+      "learning_rate": 3e-05,
+      "loss": 0.1617,
+      "step": 4726
+    },
+    {
+      "epoch": 38.75,
+      "learning_rate": 3e-05,
+      "loss": 0.1066,
+      "step": 4727
+    },
+    {
+      "epoch": 38.75,
+      "learning_rate": 3e-05,
+      "loss": 0.1553,
+      "step": 4728
+    },
+    {
+      "epoch": 38.76,
+      "learning_rate": 3e-05,
+      "loss": 0.1766,
+      "step": 4729
+    },
+    {
+      "epoch": 38.77,
+      "learning_rate": 3e-05,
+      "loss": 0.2145,
+      "step": 4730
+    },
+    {
+      "epoch": 38.78,
+      "learning_rate": 3e-05,
+      "loss": 0.199,
+      "step": 4731
+    },
+    {
+      "epoch": 38.79,
+      "learning_rate": 3e-05,
+      "loss": 0.1055,
+      "step": 4732
+    },
+    {
+      "epoch": 38.8,
+      "learning_rate": 3e-05,
+      "loss": 0.2401,
+      "step": 4733
+    },
+    {
+      "epoch": 38.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1179,
+      "step": 4734
+    },
+    {
+      "epoch": 38.81,
+      "learning_rate": 3e-05,
+      "loss": 0.1125,
+      "step": 4735
+    },
+    {
+      "epoch": 38.82,
+      "learning_rate": 3e-05,
+      "loss": 0.1311,
+      "step": 4736
+    },
+    {
+      "epoch": 38.83,
+      "learning_rate": 3e-05,
+      "loss": 0.1312,
+      "step": 4737
+    },
+    {
+      "epoch": 38.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1559,
+      "step": 4738
+    },
+    {
+      "epoch": 38.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1035,
+      "step": 4739
+    },
+    {
+      "epoch": 38.85,
+      "learning_rate": 3e-05,
+      "loss": 0.1565,
+      "step": 4740
+    },
+    {
+      "epoch": 38.86,
+      "learning_rate": 3e-05,
+      "loss": 0.1451,
+      "step": 4741
+    },
+    {
+      "epoch": 38.87,
+      "learning_rate": 3e-05,
+      "loss": 0.2727,
+      "step": 4742
+    },
+    {
+      "epoch": 38.88,
+      "learning_rate": 3e-05,
+      "loss": 0.1053,
+      "step": 4743
+    },
+    {
+      "epoch": 38.89,
+      "learning_rate": 3e-05,
+      "loss": 0.145,
+      "step": 4744
+    },
+    {
+      "epoch": 38.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1266,
+      "step": 4745
+    },
+    {
+      "epoch": 38.9,
+      "learning_rate": 3e-05,
+      "loss": 0.0977,
+      "step": 4746
+    },
+    {
+      "epoch": 38.91,
+      "learning_rate": 3e-05,
+      "loss": 0.1493,
+      "step": 4747
+    },
+    {
+      "epoch": 38.92,
+      "learning_rate": 3e-05,
+      "loss": 0.2189,
+      "step": 4748
+    },
+    {
+      "epoch": 38.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1436,
+      "step": 4749
+    },
+    {
+      "epoch": 38.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1221,
+      "step": 4750
+    },
+    {
+      "epoch": 38.94,
+      "learning_rate": 3e-05,
+      "loss": 0.1262,
+      "step": 4751
+    },
+    {
+      "epoch": 38.95,
+      "learning_rate": 3e-05,
+      "loss": 0.1423,
+      "step": 4752
+    },
+    {
+      "epoch": 38.96,
+      "learning_rate": 3e-05,
+      "loss": 0.0832,
+      "step": 4753
+    },
+    {
+      "epoch": 38.97,
+      "learning_rate": 3e-05,
+      "loss": 0.1166,
+      "step": 4754
+    },
+    {
+      "epoch": 38.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1656,
+      "step": 4755
+    },
+    {
+      "epoch": 38.98,
+      "learning_rate": 3e-05,
+      "loss": 0.3464,
+      "step": 4756
+    },
+    {
+      "epoch": 38.99,
+      "learning_rate": 3e-05,
+      "loss": 0.2224,
+      "step": 4757
+    },
+    {
+      "epoch": 39.0,
+      "learning_rate": 3e-05,
+      "loss": 0.1798,
+      "step": 4758
+    },
+    {
+      "epoch": 39.0,
+      "eval_loss": 0.260009765625,
+      "eval_mae": 0.36962890625,
+      "eval_mse": 0.260009765625,
+      "eval_runtime": 51.1326,
+      "eval_samples_per_second": 3234.592,
+      "eval_steps_per_second": 0.274,
+      "step": 4758
+    },
+    {
+      "epoch": 39.01,
+      "learning_rate": 3e-05,
+      "loss": 0.1237,
+      "step": 4759
+    },
+    {
+      "epoch": 39.02,
+      "learning_rate": 3e-05,
+      "loss": 0.0697,
+      "step": 4760
+    },
+    {
+      "epoch": 39.02,
+      "learning_rate": 3e-05,
+      "loss": 0.2603,
+      "step": 4761
+    },
+    {
+      "epoch": 39.03,
+      "learning_rate": 3e-05,
+      "loss": 0.3989,
+      "step": 4762
+    },
+    {
+      "epoch": 39.04,
+      "learning_rate": 3e-05,
+      "loss": 0.1165,
+      "step": 4763
+    },
+    {
+      "epoch": 39.05,
+      "learning_rate": 3e-05,
+      "loss": 0.1677,
+      "step": 4764
+    },
+    {
+      "epoch": 39.06,
+      "learning_rate": 3e-05,
+      "loss": 0.6787,
+      "step": 4765
+    },
+    {
+      "epoch": 39.07,
+      "learning_rate": 3e-05,
+      "loss": 0.0995,
+      "step": 4766
+    },
+    {
+      "epoch": 39.07,
+      "learning_rate": 3e-05,
+      "loss": 0.2147,
+      "step": 4767
+    },
+    {
+      "epoch": 39.08,
+      "learning_rate": 3e-05,
+      "loss": 0.1356,
+      "step": 4768
+    },
+    {
+      "epoch": 39.09,
+      "learning_rate": 3e-05,
+      "loss": 0.0938,
+      "step": 4769
+    },
+    {
+      "epoch": 39.1,
+      "learning_rate": 3e-05,
+      "loss": 0.1053,
+      "step": 4770
+    },
+    {
+      "epoch": 39.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1403,
+      "step": 4771
+    },
+    {
+      "epoch": 39.11,
+      "learning_rate": 3e-05,
+      "loss": 0.0966,
+      "step": 4772
+    },
+    {
+      "epoch": 39.12,
+      "learning_rate": 3e-05,
+      "loss": 0.0971,
+      "step": 4773
+    },
+    {
+      "epoch": 39.13,
+      "learning_rate": 3e-05,
+      "loss": 0.1337,
+      "step": 4774
+    },
+    {
+      "epoch": 39.14,
+      "learning_rate": 3e-05,
+      "loss": 0.1388,
+      "step": 4775
+    },
+    {
+      "epoch": 39.15,
+      "learning_rate": 3e-05,
+      "loss": 0.105,
+      "step": 4776
+    },
+    {
+      "epoch": 39.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1638,
+      "step": 4777
+    },
+    {
+      "epoch": 39.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1633,
+      "step": 4778
+    },
+    {
+      "epoch": 39.17,
+      "learning_rate": 3e-05,
+      "loss": 0.1135,
+      "step": 4779
+    },
+    {
+      "epoch": 39.18,
+      "learning_rate": 3e-05,
+      "loss": 0.1932,
+      "step": 4780
+    },
+    {
+      "epoch": 39.19,
+      "learning_rate": 3e-05,
+      "loss": 0.1493,
+      "step": 4781
+    },
+    {
+      "epoch": 39.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1578,
+      "step": 4782
+    },
+    {
+      "epoch": 39.2,
+      "learning_rate": 3e-05,
+      "loss": 0.0908,
+      "step": 4783
+    },
+    {
+      "epoch": 39.21,
+      "learning_rate": 3e-05,
+      "loss": 0.1744,
+      "step": 4784
+    },
+    {
+      "epoch": 39.22,
+      "learning_rate": 3e-05,
+      "loss": 0.1729,
+      "step": 4785
+    },
+    {
+      "epoch": 39.23,
+      "learning_rate": 3e-05,
+      "loss": 0.1772,
+      "step": 4786
+    },
+    {
+      "epoch": 39.24,
+      "learning_rate": 3e-05,
+      "loss": 0.1376,
+      "step": 4787
+    },
+    {
+      "epoch": 39.25,
+      "learning_rate": 3e-05,
+      "loss": 0.2391,
+      "step": 4788
+    },
+    {
+      "epoch": 39.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1968,
+      "step": 4789
+    },
+    {
+      "epoch": 39.26,
+      "learning_rate": 3e-05,
+      "loss": 0.1064,
+      "step": 4790
+    },
+    {
+      "epoch": 39.27,
+      "learning_rate": 3e-05,
+      "loss": 0.1526,
+      "step": 4791
+    },
+    {
+      "epoch": 39.28,
+      "learning_rate": 3e-05,
+      "loss": 0.1131,
+      "step": 4792
+    },
+    {
+      "epoch": 39.29,
+      "learning_rate": 3e-05,
+      "loss": 0.1345,
+      "step": 4793
+    },
+    {
+      "epoch": 39.3,
+      "learning_rate": 3e-05,
+      "loss": 0.165,
+      "step": 4794
+    },
+    {
+      "epoch": 39.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1268,
+      "step": 4795
+    },
+    {
+      "epoch": 39.31,
+      "learning_rate": 3e-05,
+      "loss": 0.0956,
+      "step": 4796
+    },
+    {
+      "epoch": 39.32,
+      "learning_rate": 3e-05,
+      "loss": 0.4219,
+      "step": 4797
+    },
+    {
+      "epoch": 39.33,
+      "learning_rate": 3e-05,
+      "loss": 0.1533,
+      "step": 4798
+    },
+    {
+      "epoch": 39.34,
+      "learning_rate": 3e-05,
+      "loss": 0.0982,
+      "step": 4799
+    },
+    {
+      "epoch": 39.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1604,
+      "step": 4800
+    },
+    {
+      "epoch": 39.35,
+      "learning_rate": 3e-05,
+      "loss": 0.1365,
+      "step": 4801
+    },
+    {
+      "epoch": 39.36,
+      "learning_rate": 3e-05,
+      "loss": 0.1494,
+      "step": 4802
+    },
+    {
+      "epoch": 39.37,
+      "learning_rate": 3e-05,
+      "loss": 0.157,
+      "step": 4803
+    },
+    {
+      "epoch": 39.38,
+      "learning_rate": 3e-05,
+      "loss": 0.1436,
+      "step": 4804
+    },
+    {
+      "epoch": 39.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1479,
+      "step": 4805
+    },
+    {
+      "epoch": 39.39,
+      "learning_rate": 3e-05,
+      "loss": 0.0614,
+      "step": 4806
+    },
+    {
+      "epoch": 39.4,
+      "learning_rate": 3e-05,
+      "loss": 0.1342,
+      "step": 4807
+    },
+    {
+      "epoch": 39.41,
+      "learning_rate": 3e-05,
+      "loss": 0.1392,
+      "step": 4808
+    },
+    {
+      "epoch": 39.42,
+      "learning_rate": 3e-05,
+      "loss": 0.1586,
+      "step": 4809
+    },
+    {
+      "epoch": 39.43,
+      "learning_rate": 3e-05,
+      "loss": 0.1633,
+      "step": 4810
+    },
+    {
+      "epoch": 39.43,
+      "learning_rate": 3e-05,
+      "loss": 0.1163,
+      "step": 4811
+    },
+    {
+      "epoch": 39.44,
+      "learning_rate": 3e-05,
+      "loss": 0.1824,
+      "step": 4812
+    },
+    {
+      "epoch": 39.45,
+      "learning_rate": 3e-05,
+      "loss": 0.09,
+      "step": 4813
+    },
+    {
+      "epoch": 39.46,
+      "learning_rate": 3e-05,
+      "loss": 0.1111,
+      "step": 4814
+    },
+    {
+      "epoch": 39.47,
+      "learning_rate": 3e-05,
+      "loss": 0.1277,
+      "step": 4815
+    },
+    {
+      "epoch": 39.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1412,
+      "step": 4816
+    },
+    {
+      "epoch": 39.48,
+      "learning_rate": 3e-05,
+      "loss": 0.2205,
+      "step": 4817
+    },
+    {
+      "epoch": 39.49,
+      "learning_rate": 3e-05,
+      "loss": 0.0983,
+      "step": 4818
+    },
+    {
+      "epoch": 39.5,
+      "learning_rate": 3e-05,
+      "loss": 0.1538,
+      "step": 4819
+    },
+    {
+      "epoch": 39.51,
+      "learning_rate": 3e-05,
+      "loss": 0.2976,
+      "step": 4820
+    },
+    {
+      "epoch": 39.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1609,
+      "step": 4821
+    },
+    {
+      "epoch": 39.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1438,
+      "step": 4822
+    },
+    {
+      "epoch": 39.53,
+      "learning_rate": 3e-05,
+      "loss": 0.097,
+      "step": 4823
+    },
+    {
+      "epoch": 39.54,
+      "learning_rate": 3e-05,
+      "loss": 0.1302,
+      "step": 4824
+    },
+    {
+      "epoch": 39.55,
+      "learning_rate": 3e-05,
+      "loss": 0.2023,
+      "step": 4825
+    },
+    {
+      "epoch": 39.56,
+      "learning_rate": 3e-05,
+      "loss": 0.0986,
+      "step": 4826
+    },
+    {
+      "epoch": 39.57,
+      "learning_rate": 3e-05,
+      "loss": 0.1449,
+      "step": 4827
+    },
+    {
+      "epoch": 39.57,
+      "learning_rate": 3e-05,
+      "loss": 0.0981,
+      "step": 4828
+    },
+    {
+      "epoch": 39.58,
+      "learning_rate": 3e-05,
+      "loss": 0.2439,
+      "step": 4829
+    },
+    {
+      "epoch": 39.59,
+      "learning_rate": 3e-05,
+      "loss": 0.1196,
+      "step": 4830
+    },
+    {
+      "epoch": 39.6,
+      "learning_rate": 3e-05,
+      "loss": 0.1293,
+      "step": 4831
+    },
+    {
+      "epoch": 39.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1047,
+      "step": 4832
+    },
+    {
+      "epoch": 39.61,
+      "learning_rate": 3e-05,
+      "loss": 0.2006,
+      "step": 4833
+    },
+    {
+      "epoch": 39.62,
+      "learning_rate": 3e-05,
+      "loss": 0.2112,
+      "step": 4834
+    },
+    {
+      "epoch": 39.63,
+      "learning_rate": 3e-05,
+      "loss": 0.1143,
+      "step": 4835
+    },
+    {
+      "epoch": 39.64,
+      "learning_rate": 3e-05,
+      "loss": 0.1271,
+      "step": 4836
+    },
+    {
+      "epoch": 39.65,
+      "learning_rate": 3e-05,
+      "loss": 0.1305,
+      "step": 4837
+    },
+    {
+      "epoch": 39.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1282,
+      "step": 4838
+    },
+    {
+      "epoch": 39.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1477,
+      "step": 4839
+    },
+    {
+      "epoch": 39.67,
+      "learning_rate": 3e-05,
+      "loss": 0.0962,
+      "step": 4840
+    },
+    {
+      "epoch": 39.68,
+      "learning_rate": 3e-05,
+      "loss": 0.1891,
+      "step": 4841
+    },
+    {
+      "epoch": 39.69,
+      "learning_rate": 3e-05,
+      "loss": 0.1893,
+      "step": 4842
+    },
+    {
+      "epoch": 39.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1299,
+      "step": 4843
+    },
+    {
+      "epoch": 39.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1209,
+      "step": 4844
+    },
+    {
+      "epoch": 39.71,
+      "learning_rate": 3e-05,
+      "loss": 0.1364,
+      "step": 4845
+    },
+    {
+      "epoch": 39.72,
+      "learning_rate": 3e-05,
+      "loss": 0.1492,
+      "step": 4846
+    },
+    {
+      "epoch": 39.73,
+      "learning_rate": 3e-05,
+      "loss": 0.1554,
+      "step": 4847
+    },
+    {
+      "epoch": 39.74,
+      "learning_rate": 3e-05,
+      "loss": 0.1294,
+      "step": 4848
+    },
+    {
+      "epoch": 39.75,
+      "learning_rate": 3e-05,
+      "loss": 0.0951,
+      "step": 4849
+    },
+    {
+      "epoch": 39.75,
+      "learning_rate": 3e-05,
+      "loss": 0.1958,
+      "step": 4850
+    },
+    {
+      "epoch": 39.76,
+      "learning_rate": 3e-05,
+      "loss": 0.0831,
+      "step": 4851
+    },
+    {
+      "epoch": 39.77,
+      "learning_rate": 3e-05,
+      "loss": 0.1188,
+      "step": 4852
+    },
+    {
+      "epoch": 39.78,
+      "learning_rate": 3e-05,
+      "loss": 0.117,
+      "step": 4853
+    },
+    {
+      "epoch": 39.79,
+      "learning_rate": 3e-05,
+      "loss": 0.1447,
+      "step": 4854
+    },
+    {
+      "epoch": 39.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1782,
+      "step": 4855
+    },
+    {
+      "epoch": 39.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1462,
+      "step": 4856
+    },
+    {
+      "epoch": 39.81,
+      "learning_rate": 3e-05,
+      "loss": 0.2393,
+      "step": 4857
+    },
+    {
+      "epoch": 39.82,
+      "learning_rate": 3e-05,
+      "loss": 0.1364,
+      "step": 4858
+    },
+    {
+      "epoch": 39.83,
+      "learning_rate": 3e-05,
+      "loss": 0.1324,
+      "step": 4859
+    },
+    {
+      "epoch": 39.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1296,
+      "step": 4860
+    },
+    {
+      "epoch": 39.84,
+      "learning_rate": 3e-05,
+      "loss": 0.167,
+      "step": 4861
+    },
+    {
+      "epoch": 39.85,
+      "learning_rate": 3e-05,
+      "loss": 0.1185,
+      "step": 4862
+    },
+    {
+      "epoch": 39.86,
+      "learning_rate": 3e-05,
+      "loss": 0.1727,
+      "step": 4863
+    },
+    {
+      "epoch": 39.87,
+      "learning_rate": 3e-05,
+      "loss": 0.1475,
+      "step": 4864
+    },
+    {
+      "epoch": 39.88,
+      "learning_rate": 3e-05,
+      "loss": 0.1451,
+      "step": 4865
+    },
+    {
+      "epoch": 39.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1692,
+      "step": 4866
+    },
+    {
+      "epoch": 39.89,
+      "learning_rate": 3e-05,
+      "loss": 0.3013,
+      "step": 4867
+    },
+    {
+      "epoch": 39.9,
+      "learning_rate": 3e-05,
+      "loss": 0.251,
+      "step": 4868
+    },
+    {
+      "epoch": 39.91,
+      "learning_rate": 3e-05,
+      "loss": 0.1824,
+      "step": 4869
+    },
+    {
+      "epoch": 39.92,
+      "learning_rate": 3e-05,
+      "loss": 0.1361,
+      "step": 4870
+    },
+    {
+      "epoch": 39.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1416,
+      "step": 4871
+    },
+    {
+      "epoch": 39.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1366,
+      "step": 4872
+    },
+    {
+      "epoch": 39.94,
+      "learning_rate": 3e-05,
+      "loss": 0.1676,
+      "step": 4873
+    },
+    {
+      "epoch": 39.95,
+      "learning_rate": 3e-05,
+      "loss": 0.2659,
+      "step": 4874
+    },
+    {
+      "epoch": 39.96,
+      "learning_rate": 3e-05,
+      "loss": 0.1511,
+      "step": 4875
+    },
+    {
+      "epoch": 39.97,
+      "learning_rate": 3e-05,
+      "loss": 0.1874,
+      "step": 4876
+    },
+    {
+      "epoch": 39.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1227,
+      "step": 4877
+    },
+    {
+      "epoch": 39.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1164,
+      "step": 4878
+    },
+    {
+      "epoch": 39.99,
+      "learning_rate": 3e-05,
+      "loss": 0.1832,
+      "step": 4879
+    },
+    {
+      "epoch": 40.0,
+      "learning_rate": 3e-05,
+      "loss": 0.039,
+      "step": 4880
+    },
+    {
+      "epoch": 40.0,
+      "eval_loss": 0.258056640625,
+      "eval_mae": 0.367431640625,
+      "eval_mse": 0.258056640625,
+      "eval_runtime": 51.1498,
+      "eval_samples_per_second": 3233.502,
+      "eval_steps_per_second": 0.274,
+      "step": 4880
+    },
+    {
+      "epoch": 40.01,
+      "learning_rate": 3e-05,
+      "loss": 0.1512,
+      "step": 4881
+    },
+    {
+      "epoch": 40.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1251,
+      "step": 4882
+    },
+    {
+      "epoch": 40.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1376,
+      "step": 4883
+    },
+    {
+      "epoch": 40.03,
+      "learning_rate": 3e-05,
+      "loss": 0.1802,
+      "step": 4884
+    },
+    {
+      "epoch": 40.04,
+      "learning_rate": 3e-05,
+      "loss": 0.1624,
+      "step": 4885
+    },
+    {
+      "epoch": 40.05,
+      "learning_rate": 3e-05,
+      "loss": 0.1158,
+      "step": 4886
+    },
+    {
+      "epoch": 40.06,
+      "learning_rate": 3e-05,
+      "loss": 0.2216,
+      "step": 4887
+    },
+    {
+      "epoch": 40.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1143,
+      "step": 4888
+    },
+    {
+      "epoch": 40.07,
+      "learning_rate": 3e-05,
+      "loss": 0.145,
+      "step": 4889
+    },
+    {
+      "epoch": 40.08,
+      "learning_rate": 3e-05,
+      "loss": 0.1074,
+      "step": 4890
+    },
+    {
+      "epoch": 40.09,
+      "learning_rate": 3e-05,
+      "loss": 0.126,
+      "step": 4891
+    },
+    {
+      "epoch": 40.1,
+      "learning_rate": 3e-05,
+      "loss": 0.1002,
+      "step": 4892
+    },
+    {
+      "epoch": 40.11,
+      "learning_rate": 3e-05,
+      "loss": 0.118,
+      "step": 4893
+    },
+    {
+      "epoch": 40.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1572,
+      "step": 4894
+    },
+    {
+      "epoch": 40.12,
+      "learning_rate": 3e-05,
+      "loss": 0.1528,
+      "step": 4895
+    },
+    {
+      "epoch": 40.13,
+      "learning_rate": 3e-05,
+      "loss": 0.1271,
+      "step": 4896
+    },
+    {
+      "epoch": 40.14,
+      "learning_rate": 3e-05,
+      "loss": 0.2235,
+      "step": 4897
+    },
+    {
+      "epoch": 40.15,
+      "learning_rate": 3e-05,
+      "loss": 0.126,
+      "step": 4898
+    },
+    {
+      "epoch": 40.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1409,
+      "step": 4899
+    },
+    {
+      "epoch": 40.16,
+      "learning_rate": 3e-05,
+      "loss": 0.105,
+      "step": 4900
+    },
+    {
+      "epoch": 40.17,
+      "learning_rate": 3e-05,
+      "loss": 0.1263,
+      "step": 4901
+    },
+    {
+      "epoch": 40.18,
+      "learning_rate": 3e-05,
+      "loss": 0.1211,
+      "step": 4902
+    },
+    {
+      "epoch": 40.19,
+      "learning_rate": 3e-05,
+      "loss": 0.1238,
+      "step": 4903
+    },
+    {
+      "epoch": 40.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1455,
+      "step": 4904
+    },
+    {
+      "epoch": 40.2,
+      "learning_rate": 3e-05,
+      "loss": 0.0977,
+      "step": 4905
+    },
+    {
+      "epoch": 40.21,
+      "learning_rate": 3e-05,
+      "loss": 0.1353,
+      "step": 4906
+    },
+    {
+      "epoch": 40.22,
+      "learning_rate": 3e-05,
+      "loss": 0.1313,
+      "step": 4907
+    },
+    {
+      "epoch": 40.23,
+      "learning_rate": 3e-05,
+      "loss": 0.0809,
+      "step": 4908
+    },
+    {
+      "epoch": 40.24,
+      "learning_rate": 3e-05,
+      "loss": 0.1359,
+      "step": 4909
+    },
+    {
+      "epoch": 40.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1459,
+      "step": 4910
+    },
+    {
+      "epoch": 40.25,
+      "learning_rate": 3e-05,
+      "loss": 0.0962,
+      "step": 4911
+    },
+    {
+      "epoch": 40.26,
+      "learning_rate": 3e-05,
+      "loss": 0.1444,
+      "step": 4912
+    },
+    {
+      "epoch": 40.27,
+      "learning_rate": 3e-05,
+      "loss": 0.1344,
+      "step": 4913
+    },
+    {
+      "epoch": 40.28,
+      "learning_rate": 3e-05,
+      "loss": 0.0823,
+      "step": 4914
+    },
+    {
+      "epoch": 40.29,
+      "learning_rate": 3e-05,
+      "loss": 0.1396,
+      "step": 4915
+    },
+    {
+      "epoch": 40.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1406,
+      "step": 4916
+    },
+    {
+      "epoch": 40.3,
+      "learning_rate": 3e-05,
+      "loss": 0.0944,
+      "step": 4917
+    },
+    {
+      "epoch": 40.31,
+      "learning_rate": 3e-05,
+      "loss": 0.12,
+      "step": 4918
+    },
+    {
+      "epoch": 40.32,
+      "learning_rate": 3e-05,
+      "loss": 0.1139,
+      "step": 4919
+    },
+    {
+      "epoch": 40.33,
+      "learning_rate": 3e-05,
+      "loss": 0.1106,
+      "step": 4920
+    },
+    {
+      "epoch": 40.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1757,
+      "step": 4921
+    },
+    {
+      "epoch": 40.34,
+      "learning_rate": 3e-05,
+      "loss": 0.2108,
+      "step": 4922
+    },
+    {
+      "epoch": 40.35,
+      "learning_rate": 3e-05,
+      "loss": 0.1475,
+      "step": 4923
+    },
+    {
+      "epoch": 40.36,
+      "learning_rate": 3e-05,
+      "loss": 0.1707,
+      "step": 4924
+    },
+    {
+      "epoch": 40.37,
+      "learning_rate": 3e-05,
+      "loss": 0.1483,
+      "step": 4925
+    },
+    {
+      "epoch": 40.38,
+      "learning_rate": 3e-05,
+      "loss": 0.1377,
+      "step": 4926
+    },
+    {
+      "epoch": 40.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1138,
+      "step": 4927
+    },
+    {
+      "epoch": 40.39,
+      "learning_rate": 3e-05,
+      "loss": 0.112,
+      "step": 4928
+    },
+    {
+      "epoch": 40.4,
+      "learning_rate": 3e-05,
+      "loss": 0.1151,
+      "step": 4929
+    },
+    {
+      "epoch": 40.41,
+      "learning_rate": 3e-05,
+      "loss": 0.1041,
+      "step": 4930
+    },
+    {
+      "epoch": 40.42,
+      "learning_rate": 3e-05,
+      "loss": 0.1063,
+      "step": 4931
+    },
+    {
+      "epoch": 40.43,
+      "learning_rate": 3e-05,
+      "loss": 0.141,
+      "step": 4932
+    },
+    {
+      "epoch": 40.43,
+      "learning_rate": 3e-05,
+      "loss": 0.1383,
+      "step": 4933
+    },
+    {
+      "epoch": 40.44,
+      "learning_rate": 3e-05,
+      "loss": 0.1022,
+      "step": 4934
+    },
+    {
+      "epoch": 40.45,
+      "learning_rate": 3e-05,
+      "loss": 0.2133,
+      "step": 4935
+    },
+    {
+      "epoch": 40.46,
+      "learning_rate": 3e-05,
+      "loss": 0.1919,
+      "step": 4936
+    },
+    {
+      "epoch": 40.47,
+      "learning_rate": 3e-05,
+      "loss": 0.1381,
+      "step": 4937
+    },
+    {
+      "epoch": 40.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1637,
+      "step": 4938
+    },
+    {
+      "epoch": 40.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1289,
+      "step": 4939
+    },
+    {
+      "epoch": 40.49,
+      "learning_rate": 3e-05,
+      "loss": 0.1288,
+      "step": 4940
+    },
+    {
+      "epoch": 40.5,
+      "learning_rate": 3e-05,
+      "loss": 0.153,
+      "step": 4941
+    },
+    {
+      "epoch": 40.51,
+      "learning_rate": 3e-05,
+      "loss": 0.1376,
+      "step": 4942
+    },
+    {
+      "epoch": 40.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1281,
+      "step": 4943
+    },
+    {
+      "epoch": 40.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1108,
+      "step": 4944
+    },
+    {
+      "epoch": 40.53,
+      "learning_rate": 3e-05,
+      "loss": 0.0737,
+      "step": 4945
+    },
+    {
+      "epoch": 40.54,
+      "learning_rate": 3e-05,
+      "loss": 0.1074,
+      "step": 4946
+    },
+    {
+      "epoch": 40.55,
+      "learning_rate": 3e-05,
+      "loss": 0.1915,
+      "step": 4947
+    },
+    {
+      "epoch": 40.56,
+      "learning_rate": 3e-05,
+      "loss": 0.1392,
+      "step": 4948
+    },
+    {
+      "epoch": 40.57,
+      "learning_rate": 3e-05,
+      "loss": 0.1324,
+      "step": 4949
+    },
+    {
+      "epoch": 40.57,
+      "learning_rate": 3e-05,
+      "loss": 0.1335,
+      "step": 4950
+    },
+    {
+      "epoch": 40.58,
+      "learning_rate": 3e-05,
+      "loss": 0.1342,
+      "step": 4951
+    },
+    {
+      "epoch": 40.59,
+      "learning_rate": 3e-05,
+      "loss": 0.1284,
+      "step": 4952
+    },
+    {
+      "epoch": 40.6,
+      "learning_rate": 3e-05,
+      "loss": 0.1335,
+      "step": 4953
+    },
+    {
+      "epoch": 40.61,
+      "learning_rate": 3e-05,
+      "loss": 0.0995,
+      "step": 4954
+    },
+    {
+      "epoch": 40.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1542,
+      "step": 4955
+    },
+    {
+      "epoch": 40.62,
+      "learning_rate": 3e-05,
+      "loss": 0.2289,
+      "step": 4956
+    },
+    {
+      "epoch": 40.63,
+      "learning_rate": 3e-05,
+      "loss": 0.101,
+      "step": 4957
+    },
+    {
+      "epoch": 40.64,
+      "learning_rate": 3e-05,
+      "loss": 0.1288,
+      "step": 4958
+    },
+    {
+      "epoch": 40.65,
+      "learning_rate": 3e-05,
+      "loss": 0.1427,
+      "step": 4959
+    },
+    {
+      "epoch": 40.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1564,
+      "step": 4960
+    },
+    {
+      "epoch": 40.66,
+      "learning_rate": 3e-05,
+      "loss": 0.0947,
+      "step": 4961
+    },
+    {
+      "epoch": 40.67,
+      "learning_rate": 3e-05,
+      "loss": 0.1757,
+      "step": 4962
+    },
+    {
+      "epoch": 40.68,
+      "learning_rate": 3e-05,
+      "loss": 0.1262,
+      "step": 4963
+    },
+    {
+      "epoch": 40.69,
+      "learning_rate": 3e-05,
+      "loss": 0.1132,
+      "step": 4964
+    },
+    {
+      "epoch": 40.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1085,
+      "step": 4965
+    },
+    {
+      "epoch": 40.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1318,
+      "step": 4966
+    },
+    {
+      "epoch": 40.71,
+      "learning_rate": 3e-05,
+      "loss": 0.3977,
+      "step": 4967
+    },
+    {
+      "epoch": 40.72,
+      "learning_rate": 3e-05,
+      "loss": 0.208,
+      "step": 4968
+    },
+    {
+      "epoch": 40.73,
+      "learning_rate": 3e-05,
+      "loss": 0.1927,
+      "step": 4969
+    },
+    {
+      "epoch": 40.74,
+      "learning_rate": 3e-05,
+      "loss": 0.2434,
+      "step": 4970
+    },
+    {
+      "epoch": 40.75,
+      "learning_rate": 3e-05,
+      "loss": 0.1691,
+      "step": 4971
+    },
+    {
+      "epoch": 40.75,
+      "learning_rate": 3e-05,
+      "loss": 0.103,
+      "step": 4972
+    },
+    {
+      "epoch": 40.76,
+      "learning_rate": 3e-05,
+      "loss": 0.1783,
+      "step": 4973
+    },
+    {
+      "epoch": 40.77,
+      "learning_rate": 3e-05,
+      "loss": 0.1135,
+      "step": 4974
+    },
+    {
+      "epoch": 40.78,
+      "learning_rate": 3e-05,
+      "loss": 0.1737,
+      "step": 4975
+    },
+    {
+      "epoch": 40.79,
+      "learning_rate": 3e-05,
+      "loss": 0.099,
+      "step": 4976
+    },
+    {
+      "epoch": 40.8,
+      "learning_rate": 3e-05,
+      "loss": 0.0872,
+      "step": 4977
+    },
+    {
+      "epoch": 40.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1064,
+      "step": 4978
+    },
+    {
+      "epoch": 40.81,
+      "learning_rate": 3e-05,
+      "loss": 0.1758,
+      "step": 4979
+    },
+    {
+      "epoch": 40.82,
+      "learning_rate": 3e-05,
+      "loss": 0.1361,
+      "step": 4980
+    },
+    {
+      "epoch": 40.83,
+      "learning_rate": 3e-05,
+      "loss": 0.1388,
+      "step": 4981
+    },
+    {
+      "epoch": 40.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1204,
+      "step": 4982
+    },
+    {
+      "epoch": 40.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1057,
+      "step": 4983
+    },
+    {
+      "epoch": 40.85,
+      "learning_rate": 3e-05,
+      "loss": 0.1021,
+      "step": 4984
+    },
+    {
+      "epoch": 40.86,
+      "learning_rate": 3e-05,
+      "loss": 0.1447,
+      "step": 4985
+    },
+    {
+      "epoch": 40.87,
+      "learning_rate": 3e-05,
+      "loss": 0.0926,
+      "step": 4986
+    },
+    {
+      "epoch": 40.88,
+      "learning_rate": 3e-05,
+      "loss": 0.1146,
+      "step": 4987
+    },
+    {
+      "epoch": 40.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1289,
+      "step": 4988
+    },
+    {
+      "epoch": 40.89,
+      "learning_rate": 3e-05,
+      "loss": 0.218,
+      "step": 4989
+    },
+    {
+      "epoch": 40.9,
+      "learning_rate": 3e-05,
+      "loss": 0.1395,
+      "step": 4990
+    },
+    {
+      "epoch": 40.91,
+      "learning_rate": 3e-05,
+      "loss": 0.2002,
+      "step": 4991
+    },
+    {
+      "epoch": 40.92,
+      "learning_rate": 3e-05,
+      "loss": 0.1841,
+      "step": 4992
+    },
+    {
+      "epoch": 40.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1874,
+      "step": 4993
+    },
+    {
+      "epoch": 40.93,
+      "learning_rate": 3e-05,
+      "loss": 0.0922,
+      "step": 4994
+    },
+    {
+      "epoch": 40.94,
+      "learning_rate": 3e-05,
+      "loss": 0.1633,
+      "step": 4995
+    },
+    {
+      "epoch": 40.95,
+      "learning_rate": 3e-05,
+      "loss": 0.1877,
+      "step": 4996
+    },
+    {
+      "epoch": 40.96,
+      "learning_rate": 3e-05,
+      "loss": 0.0947,
+      "step": 4997
+    },
+    {
+      "epoch": 40.97,
+      "learning_rate": 3e-05,
+      "loss": 0.1768,
+      "step": 4998
+    },
+    {
+      "epoch": 40.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1693,
+      "step": 4999
+    },
+    {
+      "epoch": 40.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1201,
+      "step": 5000
+    },
+    {
+      "epoch": 40.99,
+      "learning_rate": 3e-05,
+      "loss": 0.1656,
+      "step": 5001
+    },
+    {
+      "epoch": 41.0,
+      "learning_rate": 3e-05,
+      "loss": 0.2722,
+      "step": 5002
+    },
+    {
+      "epoch": 41.0,
+      "eval_loss": 0.25927734375,
+      "eval_mae": 0.367431640625,
+      "eval_mse": 0.25927734375,
+      "eval_runtime": 51.1708,
+      "eval_samples_per_second": 3232.178,
+      "eval_steps_per_second": 0.274,
+      "step": 5002
+    },
+    {
+      "epoch": 41.01,
+      "learning_rate": 3e-05,
+      "loss": 0.1267,
+      "step": 5003
+    },
+    {
+      "epoch": 41.02,
+      "learning_rate": 3e-05,
+      "loss": 0.0928,
+      "step": 5004
+    },
+    {
+      "epoch": 41.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1356,
+      "step": 5005
+    },
+    {
+      "epoch": 41.03,
+      "learning_rate": 3e-05,
+      "loss": 0.1447,
+      "step": 5006
+    },
+    {
+      "epoch": 41.04,
+      "learning_rate": 3e-05,
+      "loss": 0.1581,
+      "step": 5007
+    },
+    {
+      "epoch": 41.05,
+      "learning_rate": 3e-05,
+      "loss": 0.1163,
+      "step": 5008
+    },
+    {
+      "epoch": 41.06,
+      "learning_rate": 3e-05,
+      "loss": 0.131,
+      "step": 5009
+    },
+    {
+      "epoch": 41.07,
+      "learning_rate": 3e-05,
+      "loss": 0.2173,
+      "step": 5010
+    },
+    {
+      "epoch": 41.07,
+      "learning_rate": 3e-05,
+      "loss": 0.118,
+      "step": 5011
+    },
+    {
+      "epoch": 41.08,
+      "learning_rate": 3e-05,
+      "loss": 0.1364,
+      "step": 5012
+    },
+    {
+      "epoch": 41.09,
+      "learning_rate": 3e-05,
+      "loss": 0.1409,
+      "step": 5013
+    },
+    {
+      "epoch": 41.1,
+      "learning_rate": 3e-05,
+      "loss": 0.1338,
+      "step": 5014
+    },
+    {
+      "epoch": 41.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1367,
+      "step": 5015
+    },
+    {
+      "epoch": 41.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1615,
+      "step": 5016
+    },
+    {
+      "epoch": 41.12,
+      "learning_rate": 3e-05,
+      "loss": 0.13,
+      "step": 5017
+    },
+    {
+      "epoch": 41.13,
+      "learning_rate": 3e-05,
+      "loss": 0.1414,
+      "step": 5018
+    },
+    {
+      "epoch": 41.14,
+      "learning_rate": 3e-05,
+      "loss": 0.0935,
+      "step": 5019
+    },
+    {
+      "epoch": 41.15,
+      "learning_rate": 3e-05,
+      "loss": 0.1503,
+      "step": 5020
+    },
+    {
+      "epoch": 41.16,
+      "learning_rate": 3e-05,
+      "loss": 0.2036,
+      "step": 5021
+    },
+    {
+      "epoch": 41.16,
+      "learning_rate": 3e-05,
+      "loss": 0.176,
+      "step": 5022
+    },
+    {
+      "epoch": 41.17,
+      "learning_rate": 3e-05,
+      "loss": 0.1805,
+      "step": 5023
+    },
+    {
+      "epoch": 41.18,
+      "learning_rate": 3e-05,
+      "loss": 0.1466,
+      "step": 5024
+    },
+    {
+      "epoch": 41.19,
+      "learning_rate": 3e-05,
+      "loss": 0.1273,
+      "step": 5025
+    },
+    {
+      "epoch": 41.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1357,
+      "step": 5026
+    },
+    {
+      "epoch": 41.2,
+      "learning_rate": 3e-05,
+      "loss": 0.158,
+      "step": 5027
+    },
+    {
+      "epoch": 41.21,
+      "learning_rate": 3e-05,
+      "loss": 0.2131,
+      "step": 5028
+    },
+    {
+      "epoch": 41.22,
+      "learning_rate": 3e-05,
+      "loss": 0.0977,
+      "step": 5029
+    },
+    {
+      "epoch": 41.23,
+      "learning_rate": 3e-05,
+      "loss": 0.0965,
+      "step": 5030
+    },
+    {
+      "epoch": 41.24,
+      "learning_rate": 3e-05,
+      "loss": 0.149,
+      "step": 5031
+    },
+    {
+      "epoch": 41.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1361,
+      "step": 5032
+    },
+    {
+      "epoch": 41.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1259,
+      "step": 5033
+    },
+    {
+      "epoch": 41.26,
+      "learning_rate": 3e-05,
+      "loss": 0.1028,
+      "step": 5034
+    },
+    {
+      "epoch": 41.27,
+      "learning_rate": 3e-05,
+      "loss": 0.166,
+      "step": 5035
+    },
+    {
+      "epoch": 41.28,
+      "learning_rate": 3e-05,
+      "loss": 0.1907,
+      "step": 5036
+    },
+    {
+      "epoch": 41.29,
+      "learning_rate": 3e-05,
+      "loss": 0.1055,
+      "step": 5037
+    },
+    {
+      "epoch": 41.3,
+      "learning_rate": 3e-05,
+      "loss": 0.0964,
+      "step": 5038
+    },
+    {
+      "epoch": 41.3,
+      "learning_rate": 3e-05,
+      "loss": 0.114,
+      "step": 5039
+    },
+    {
+      "epoch": 41.31,
+      "learning_rate": 3e-05,
+      "loss": 0.135,
+      "step": 5040
+    },
+    {
+      "epoch": 41.32,
+      "learning_rate": 3e-05,
+      "loss": 0.0739,
+      "step": 5041
+    },
+    {
+      "epoch": 41.33,
+      "learning_rate": 3e-05,
+      "loss": 0.13,
+      "step": 5042
+    },
+    {
+      "epoch": 41.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1909,
+      "step": 5043
+    },
+    {
+      "epoch": 41.34,
+      "learning_rate": 3e-05,
+      "loss": 0.0938,
+      "step": 5044
+    },
+    {
+      "epoch": 41.35,
+      "learning_rate": 3e-05,
+      "loss": 0.1567,
+      "step": 5045
+    },
+    {
+      "epoch": 41.36,
+      "learning_rate": 3e-05,
+      "loss": 0.1611,
+      "step": 5046
+    },
+    {
+      "epoch": 41.37,
+      "learning_rate": 3e-05,
+      "loss": 0.1072,
+      "step": 5047
+    },
+    {
+      "epoch": 41.38,
+      "learning_rate": 3e-05,
+      "loss": 0.1266,
+      "step": 5048
+    },
+    {
+      "epoch": 41.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1526,
+      "step": 5049
+    },
+    {
+      "epoch": 41.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1458,
+      "step": 5050
+    },
+    {
+      "epoch": 41.4,
+      "learning_rate": 3e-05,
+      "loss": 0.0961,
+      "step": 5051
+    },
+    {
+      "epoch": 41.41,
+      "learning_rate": 3e-05,
+      "loss": 0.1677,
+      "step": 5052
+    },
+    {
+      "epoch": 41.42,
+      "learning_rate": 3e-05,
+      "loss": 0.1376,
+      "step": 5053
+    },
+    {
+      "epoch": 41.43,
+      "learning_rate": 3e-05,
+      "loss": 0.1707,
+      "step": 5054
+    },
+    {
+      "epoch": 41.43,
+      "learning_rate": 3e-05,
+      "loss": 0.1206,
+      "step": 5055
+    },
+    {
+      "epoch": 41.44,
+      "learning_rate": 3e-05,
+      "loss": 0.1324,
+      "step": 5056
+    },
+    {
+      "epoch": 41.45,
+      "learning_rate": 3e-05,
+      "loss": 0.2498,
+      "step": 5057
+    },
+    {
+      "epoch": 41.46,
+      "learning_rate": 3e-05,
+      "loss": 0.1802,
+      "step": 5058
+    },
+    {
+      "epoch": 41.47,
+      "learning_rate": 3e-05,
+      "loss": 0.1711,
+      "step": 5059
+    },
+    {
+      "epoch": 41.48,
+      "learning_rate": 3e-05,
+      "loss": 0.2366,
+      "step": 5060
+    },
+    {
+      "epoch": 41.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1454,
+      "step": 5061
+    },
+    {
+      "epoch": 41.49,
+      "learning_rate": 3e-05,
+      "loss": 0.0905,
+      "step": 5062
+    },
+    {
+      "epoch": 41.5,
+      "learning_rate": 3e-05,
+      "loss": 0.1047,
+      "step": 5063
+    },
+    {
+      "epoch": 41.51,
+      "learning_rate": 3e-05,
+      "loss": 0.2993,
+      "step": 5064
+    },
+    {
+      "epoch": 41.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1456,
+      "step": 5065
+    },
+    {
+      "epoch": 41.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1135,
+      "step": 5066
+    },
+    {
+      "epoch": 41.53,
+      "learning_rate": 3e-05,
+      "loss": 0.1519,
+      "step": 5067
+    },
+    {
+      "epoch": 41.54,
+      "learning_rate": 3e-05,
+      "loss": 0.1754,
+      "step": 5068
+    },
+    {
+      "epoch": 41.55,
+      "learning_rate": 3e-05,
+      "loss": 0.1871,
+      "step": 5069
+    },
+    {
+      "epoch": 41.56,
+      "learning_rate": 3e-05,
+      "loss": 0.1553,
+      "step": 5070
+    },
+    {
+      "epoch": 41.57,
+      "learning_rate": 3e-05,
+      "loss": 0.1017,
+      "step": 5071
+    },
+    {
+      "epoch": 41.57,
+      "learning_rate": 3e-05,
+      "loss": 0.1971,
+      "step": 5072
+    },
+    {
+      "epoch": 41.58,
+      "learning_rate": 3e-05,
+      "loss": 0.1821,
+      "step": 5073
+    },
+    {
+      "epoch": 41.59,
+      "learning_rate": 3e-05,
+      "loss": 0.1344,
+      "step": 5074
+    },
+    {
+      "epoch": 41.6,
+      "learning_rate": 3e-05,
+      "loss": 0.1327,
+      "step": 5075
+    },
+    {
+      "epoch": 41.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1086,
+      "step": 5076
+    },
+    {
+      "epoch": 41.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1691,
+      "step": 5077
+    },
+    {
+      "epoch": 41.62,
+      "learning_rate": 3e-05,
+      "loss": 0.1671,
+      "step": 5078
+    },
+    {
+      "epoch": 41.63,
+      "learning_rate": 3e-05,
+      "loss": 0.1681,
+      "step": 5079
+    },
+    {
+      "epoch": 41.64,
+      "learning_rate": 3e-05,
+      "loss": 0.1439,
+      "step": 5080
+    },
+    {
+      "epoch": 41.65,
+      "learning_rate": 3e-05,
+      "loss": 0.1126,
+      "step": 5081
+    },
+    {
+      "epoch": 41.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1285,
+      "step": 5082
+    },
+    {
+      "epoch": 41.66,
+      "learning_rate": 3e-05,
+      "loss": 0.0972,
+      "step": 5083
+    },
+    {
+      "epoch": 41.67,
+      "learning_rate": 3e-05,
+      "loss": 0.1652,
+      "step": 5084
+    },
+    {
+      "epoch": 41.68,
+      "learning_rate": 3e-05,
+      "loss": 0.1137,
+      "step": 5085
+    },
+    {
+      "epoch": 41.69,
+      "learning_rate": 3e-05,
+      "loss": 0.097,
+      "step": 5086
+    },
+    {
+      "epoch": 41.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1727,
+      "step": 5087
+    },
+    {
+      "epoch": 41.7,
+      "learning_rate": 3e-05,
+      "loss": 0.2202,
+      "step": 5088
+    },
+    {
+      "epoch": 41.71,
+      "learning_rate": 3e-05,
+      "loss": 0.162,
+      "step": 5089
+    },
+    {
+      "epoch": 41.72,
+      "learning_rate": 3e-05,
+      "loss": 0.1416,
+      "step": 5090
+    },
+    {
+      "epoch": 41.73,
+      "learning_rate": 3e-05,
+      "loss": 0.2014,
+      "step": 5091
+    },
+    {
+      "epoch": 41.74,
+      "learning_rate": 3e-05,
+      "loss": 0.0929,
+      "step": 5092
+    },
+    {
+      "epoch": 41.75,
+      "learning_rate": 3e-05,
+      "loss": 0.1117,
+      "step": 5093
+    },
+    {
+      "epoch": 41.75,
+      "learning_rate": 3e-05,
+      "loss": 0.136,
+      "step": 5094
+    },
+    {
+      "epoch": 41.76,
+      "learning_rate": 3e-05,
+      "loss": 0.1694,
+      "step": 5095
+    },
+    {
+      "epoch": 41.77,
+      "learning_rate": 3e-05,
+      "loss": 0.1078,
+      "step": 5096
+    },
+    {
+      "epoch": 41.78,
+      "learning_rate": 3e-05,
+      "loss": 0.1333,
+      "step": 5097
+    },
+    {
+      "epoch": 41.79,
+      "learning_rate": 3e-05,
+      "loss": 0.1227,
+      "step": 5098
+    },
+    {
+      "epoch": 41.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1935,
+      "step": 5099
+    },
+    {
+      "epoch": 41.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1724,
+      "step": 5100
+    },
+    {
+      "epoch": 41.81,
+      "learning_rate": 3e-05,
+      "loss": 0.1047,
+      "step": 5101
+    },
+    {
+      "epoch": 41.82,
+      "learning_rate": 3e-05,
+      "loss": 0.1693,
+      "step": 5102
+    },
+    {
+      "epoch": 41.83,
+      "learning_rate": 3e-05,
+      "loss": 0.1439,
+      "step": 5103
+    },
+    {
+      "epoch": 41.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1263,
+      "step": 5104
+    },
+    {
+      "epoch": 41.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1028,
+      "step": 5105
+    },
+    {
+      "epoch": 41.85,
+      "learning_rate": 3e-05,
+      "loss": 0.1809,
+      "step": 5106
+    },
+    {
+      "epoch": 41.86,
+      "learning_rate": 3e-05,
+      "loss": 0.142,
+      "step": 5107
+    },
+    {
+      "epoch": 41.87,
+      "learning_rate": 3e-05,
+      "loss": 0.13,
+      "step": 5108
+    },
+    {
+      "epoch": 41.88,
+      "learning_rate": 3e-05,
+      "loss": 0.1486,
+      "step": 5109
+    },
+    {
+      "epoch": 41.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1622,
+      "step": 5110
+    },
+    {
+      "epoch": 41.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1562,
+      "step": 5111
+    },
+    {
+      "epoch": 41.9,
+      "learning_rate": 3e-05,
+      "loss": 0.1387,
+      "step": 5112
+    },
+    {
+      "epoch": 41.91,
+      "learning_rate": 3e-05,
+      "loss": 0.1605,
+      "step": 5113
+    },
+    {
+      "epoch": 41.92,
+      "learning_rate": 3e-05,
+      "loss": 0.1433,
+      "step": 5114
+    },
+    {
+      "epoch": 41.93,
+      "learning_rate": 3e-05,
+      "loss": 0.095,
+      "step": 5115
+    },
+    {
+      "epoch": 41.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1298,
+      "step": 5116
+    },
+    {
+      "epoch": 41.94,
+      "learning_rate": 3e-05,
+      "loss": 0.1326,
+      "step": 5117
+    },
+    {
+      "epoch": 41.95,
+      "learning_rate": 3e-05,
+      "loss": 0.1796,
+      "step": 5118
+    },
+    {
+      "epoch": 41.96,
+      "learning_rate": 3e-05,
+      "loss": 0.2064,
+      "step": 5119
+    },
+    {
+      "epoch": 41.97,
+      "learning_rate": 3e-05,
+      "loss": 0.136,
+      "step": 5120
+    },
+    {
+      "epoch": 41.98,
+      "learning_rate": 3e-05,
+      "loss": 0.134,
+      "step": 5121
+    },
+    {
+      "epoch": 41.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1233,
+      "step": 5122
+    },
+    {
+      "epoch": 41.99,
+      "learning_rate": 3e-05,
+      "loss": 0.1844,
+      "step": 5123
+    },
+    {
+      "epoch": 42.0,
+      "learning_rate": 3e-05,
+      "loss": 0.3123,
+      "step": 5124
+    },
+    {
+      "epoch": 42.0,
+      "eval_loss": 0.25634765625,
+      "eval_mae": 0.365234375,
+      "eval_mse": 0.25634765625,
+      "eval_runtime": 51.2529,
+      "eval_samples_per_second": 3226.999,
+      "eval_steps_per_second": 0.273,
+      "step": 5124
+    },
+    {
+      "epoch": 42.01,
+      "learning_rate": 3e-05,
+      "loss": 0.1217,
+      "step": 5125
+    },
+    {
+      "epoch": 42.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1497,
+      "step": 5126
+    },
+    {
+      "epoch": 42.02,
+      "learning_rate": 3e-05,
+      "loss": 0.2114,
+      "step": 5127
+    },
+    {
+      "epoch": 42.03,
+      "learning_rate": 3e-05,
+      "loss": 0.3662,
+      "step": 5128
+    },
+    {
+      "epoch": 42.04,
+      "learning_rate": 3e-05,
+      "loss": 0.0924,
+      "step": 5129
+    },
+    {
+      "epoch": 42.05,
+      "learning_rate": 3e-05,
+      "loss": 0.1256,
+      "step": 5130
+    },
+    {
+      "epoch": 42.06,
+      "learning_rate": 3e-05,
+      "loss": 0.1646,
+      "step": 5131
+    },
+    {
+      "epoch": 42.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1289,
+      "step": 5132
+    },
+    {
+      "epoch": 42.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1675,
+      "step": 5133
+    },
+    {
+      "epoch": 42.08,
+      "learning_rate": 3e-05,
+      "loss": 0.2351,
+      "step": 5134
+    },
+    {
+      "epoch": 42.09,
+      "learning_rate": 3e-05,
+      "loss": 0.108,
+      "step": 5135
+    },
+    {
+      "epoch": 42.1,
+      "learning_rate": 3e-05,
+      "loss": 0.1532,
+      "step": 5136
+    },
+    {
+      "epoch": 42.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1488,
+      "step": 5137
+    },
+    {
+      "epoch": 42.11,
+      "learning_rate": 3e-05,
+      "loss": 0.131,
+      "step": 5138
+    },
+    {
+      "epoch": 42.12,
+      "learning_rate": 3e-05,
+      "loss": 0.1498,
+      "step": 5139
+    },
+    {
+      "epoch": 42.13,
+      "learning_rate": 3e-05,
+      "loss": 0.2178,
+      "step": 5140
+    },
+    {
+      "epoch": 42.14,
+      "learning_rate": 3e-05,
+      "loss": 0.1464,
+      "step": 5141
+    },
+    {
+      "epoch": 42.15,
+      "learning_rate": 3e-05,
+      "loss": 0.1696,
+      "step": 5142
+    },
+    {
+      "epoch": 42.16,
+      "learning_rate": 3e-05,
+      "loss": 0.174,
+      "step": 5143
+    },
+    {
+      "epoch": 42.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1536,
+      "step": 5144
+    },
+    {
+      "epoch": 42.17,
+      "learning_rate": 3e-05,
+      "loss": 0.1536,
+      "step": 5145
+    },
+    {
+      "epoch": 42.18,
+      "learning_rate": 3e-05,
+      "loss": 0.1506,
+      "step": 5146
+    },
+    {
+      "epoch": 42.19,
+      "learning_rate": 3e-05,
+      "loss": 0.2294,
+      "step": 5147
+    },
+    {
+      "epoch": 42.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1301,
+      "step": 5148
+    },
+    {
+      "epoch": 42.2,
+      "learning_rate": 3e-05,
+      "loss": 0.105,
+      "step": 5149
+    },
+    {
+      "epoch": 42.21,
+      "learning_rate": 3e-05,
+      "loss": 0.0876,
+      "step": 5150
+    },
+    {
+      "epoch": 42.22,
+      "learning_rate": 3e-05,
+      "loss": 0.1412,
+      "step": 5151
+    },
+    {
+      "epoch": 42.23,
+      "learning_rate": 3e-05,
+      "loss": 0.1522,
+      "step": 5152
+    },
+    {
+      "epoch": 42.24,
+      "learning_rate": 3e-05,
+      "loss": 0.2712,
+      "step": 5153
+    },
+    {
+      "epoch": 42.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1166,
+      "step": 5154
+    },
+    {
+      "epoch": 42.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1039,
+      "step": 5155
+    },
+    {
+      "epoch": 42.26,
+      "learning_rate": 3e-05,
+      "loss": 0.1263,
+      "step": 5156
+    },
+    {
+      "epoch": 42.27,
+      "learning_rate": 3e-05,
+      "loss": 0.1244,
+      "step": 5157
+    },
+    {
+      "epoch": 42.28,
+      "learning_rate": 3e-05,
+      "loss": 0.1418,
+      "step": 5158
+    },
+    {
+      "epoch": 42.29,
+      "learning_rate": 3e-05,
+      "loss": 0.108,
+      "step": 5159
+    },
+    {
+      "epoch": 42.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1437,
+      "step": 5160
+    },
+    {
+      "epoch": 42.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1294,
+      "step": 5161
+    },
+    {
+      "epoch": 42.31,
+      "learning_rate": 3e-05,
+      "loss": 0.136,
+      "step": 5162
+    },
+    {
+      "epoch": 42.32,
+      "learning_rate": 3e-05,
+      "loss": 0.0973,
+      "step": 5163
+    },
+    {
+      "epoch": 42.33,
+      "learning_rate": 3e-05,
+      "loss": 0.1248,
+      "step": 5164
+    },
+    {
+      "epoch": 42.34,
+      "learning_rate": 3e-05,
+      "loss": 0.2803,
+      "step": 5165
+    },
+    {
+      "epoch": 42.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1212,
+      "step": 5166
+    },
+    {
+      "epoch": 42.35,
+      "learning_rate": 3e-05,
+      "loss": 0.1772,
+      "step": 5167
+    },
+    {
+      "epoch": 42.36,
+      "learning_rate": 3e-05,
+      "loss": 0.14,
+      "step": 5168
+    },
+    {
+      "epoch": 42.37,
+      "learning_rate": 3e-05,
+      "loss": 0.1365,
+      "step": 5169
+    },
+    {
+      "epoch": 42.38,
+      "learning_rate": 3e-05,
+      "loss": 0.1801,
+      "step": 5170
+    },
+    {
+      "epoch": 42.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1508,
+      "step": 5171
+    },
+    {
+      "epoch": 42.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1628,
+      "step": 5172
+    },
+    {
+      "epoch": 42.4,
+      "learning_rate": 3e-05,
+      "loss": 0.1997,
+      "step": 5173
+    },
+    {
+      "epoch": 42.41,
+      "learning_rate": 3e-05,
+      "loss": 0.1779,
+      "step": 5174
+    },
+    {
+      "epoch": 42.42,
+      "learning_rate": 3e-05,
+      "loss": 0.1016,
+      "step": 5175
+    },
+    {
+      "epoch": 42.43,
+      "learning_rate": 3e-05,
+      "loss": 0.1141,
+      "step": 5176
+    },
+    {
+      "epoch": 42.43,
+      "learning_rate": 3e-05,
+      "loss": 0.1238,
+      "step": 5177
+    },
+    {
+      "epoch": 42.44,
+      "learning_rate": 3e-05,
+      "loss": 0.1071,
+      "step": 5178
+    },
+    {
+      "epoch": 42.45,
+      "learning_rate": 3e-05,
+      "loss": 0.1356,
+      "step": 5179
+    },
+    {
+      "epoch": 42.46,
+      "learning_rate": 3e-05,
+      "loss": 0.0984,
+      "step": 5180
+    },
+    {
+      "epoch": 42.47,
+      "learning_rate": 3e-05,
+      "loss": 0.153,
+      "step": 5181
+    },
+    {
+      "epoch": 42.48,
+      "learning_rate": 3e-05,
+      "loss": 0.134,
+      "step": 5182
+    },
+    {
+      "epoch": 42.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1595,
+      "step": 5183
+    },
+    {
+      "epoch": 42.49,
+      "learning_rate": 3e-05,
+      "loss": 0.1729,
+      "step": 5184
+    },
+    {
+      "epoch": 42.5,
+      "learning_rate": 3e-05,
+      "loss": 0.2264,
+      "step": 5185
+    },
+    {
+      "epoch": 42.51,
+      "learning_rate": 3e-05,
+      "loss": 0.1914,
+      "step": 5186
+    },
+    {
+      "epoch": 42.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1429,
+      "step": 5187
+    },
+    {
+      "epoch": 42.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1415,
+      "step": 5188
+    },
+    {
+      "epoch": 42.53,
+      "learning_rate": 3e-05,
+      "loss": 0.468,
+      "step": 5189
+    },
+    {
+      "epoch": 42.54,
+      "learning_rate": 3e-05,
+      "loss": 0.1132,
+      "step": 5190
+    },
+    {
+      "epoch": 42.55,
+      "learning_rate": 3e-05,
+      "loss": 0.073,
+      "step": 5191
+    },
+    {
+      "epoch": 42.56,
+      "learning_rate": 3e-05,
+      "loss": 0.0977,
+      "step": 5192
+    },
+    {
+      "epoch": 42.57,
+      "learning_rate": 3e-05,
+      "loss": 0.1249,
+      "step": 5193
+    },
+    {
+      "epoch": 42.57,
+      "learning_rate": 3e-05,
+      "loss": 0.1383,
+      "step": 5194
+    },
+    {
+      "epoch": 42.58,
+      "learning_rate": 3e-05,
+      "loss": 0.1246,
+      "step": 5195
+    },
+    {
+      "epoch": 42.59,
+      "learning_rate": 3e-05,
+      "loss": 0.1127,
+      "step": 5196
+    },
+    {
+      "epoch": 42.6,
+      "learning_rate": 3e-05,
+      "loss": 0.1259,
+      "step": 5197
+    },
+    {
+      "epoch": 42.61,
+      "learning_rate": 3e-05,
+      "loss": 0.144,
+      "step": 5198
+    },
+    {
+      "epoch": 42.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1304,
+      "step": 5199
+    },
+    {
+      "epoch": 42.62,
+      "learning_rate": 3e-05,
+      "loss": 0.1492,
+      "step": 5200
+    },
+    {
+      "epoch": 42.63,
+      "learning_rate": 3e-05,
+      "loss": 0.1334,
+      "step": 5201
+    },
+    {
+      "epoch": 42.64,
+      "learning_rate": 3e-05,
+      "loss": 0.1588,
+      "step": 5202
+    },
+    {
+      "epoch": 42.65,
+      "learning_rate": 3e-05,
+      "loss": 0.0828,
+      "step": 5203
+    },
+    {
+      "epoch": 42.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1009,
+      "step": 5204
+    },
+    {
+      "epoch": 42.66,
+      "learning_rate": 3e-05,
+      "loss": 0.0938,
+      "step": 5205
+    },
+    {
+      "epoch": 42.67,
+      "learning_rate": 3e-05,
+      "loss": 0.1073,
+      "step": 5206
+    },
+    {
+      "epoch": 42.68,
+      "learning_rate": 3e-05,
+      "loss": 0.1558,
+      "step": 5207
+    },
+    {
+      "epoch": 42.69,
+      "learning_rate": 3e-05,
+      "loss": 0.2668,
+      "step": 5208
+    },
+    {
+      "epoch": 42.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1744,
+      "step": 5209
+    },
+    {
+      "epoch": 42.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1587,
+      "step": 5210
+    },
+    {
+      "epoch": 42.71,
+      "learning_rate": 3e-05,
+      "loss": 0.1273,
+      "step": 5211
+    },
+    {
+      "epoch": 42.72,
+      "learning_rate": 3e-05,
+      "loss": 0.262,
+      "step": 5212
+    },
+    {
+      "epoch": 42.73,
+      "learning_rate": 3e-05,
+      "loss": 0.1227,
+      "step": 5213
+    },
+    {
+      "epoch": 42.74,
+      "learning_rate": 3e-05,
+      "loss": 0.153,
+      "step": 5214
+    },
+    {
+      "epoch": 42.75,
+      "learning_rate": 3e-05,
+      "loss": 0.1569,
+      "step": 5215
+    },
+    {
+      "epoch": 42.75,
+      "learning_rate": 3e-05,
+      "loss": 0.1525,
+      "step": 5216
+    },
+    {
+      "epoch": 42.76,
+      "learning_rate": 3e-05,
+      "loss": 0.1653,
+      "step": 5217
+    },
+    {
+      "epoch": 42.77,
+      "learning_rate": 3e-05,
+      "loss": 0.1249,
+      "step": 5218
+    },
+    {
+      "epoch": 42.78,
+      "learning_rate": 3e-05,
+      "loss": 0.1105,
+      "step": 5219
+    },
+    {
+      "epoch": 42.79,
+      "learning_rate": 3e-05,
+      "loss": 0.1321,
+      "step": 5220
+    },
+    {
+      "epoch": 42.8,
+      "learning_rate": 3e-05,
+      "loss": 0.2197,
+      "step": 5221
+    },
+    {
+      "epoch": 42.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1148,
+      "step": 5222
+    },
+    {
+      "epoch": 42.81,
+      "learning_rate": 3e-05,
+      "loss": 0.1833,
+      "step": 5223
+    },
+    {
+      "epoch": 42.82,
+      "learning_rate": 3e-05,
+      "loss": 0.1442,
+      "step": 5224
+    },
+    {
+      "epoch": 42.83,
+      "learning_rate": 3e-05,
+      "loss": 0.1384,
+      "step": 5225
+    },
+    {
+      "epoch": 42.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1127,
+      "step": 5226
+    },
+    {
+      "epoch": 42.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1411,
+      "step": 5227
+    },
+    {
+      "epoch": 42.85,
+      "learning_rate": 3e-05,
+      "loss": 0.1337,
+      "step": 5228
+    },
+    {
+      "epoch": 42.86,
+      "learning_rate": 3e-05,
+      "loss": 0.1326,
+      "step": 5229
+    },
+    {
+      "epoch": 42.87,
+      "learning_rate": 3e-05,
+      "loss": 0.1896,
+      "step": 5230
+    },
+    {
+      "epoch": 42.88,
+      "learning_rate": 3e-05,
+      "loss": 0.1512,
+      "step": 5231
+    },
+    {
+      "epoch": 42.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1423,
+      "step": 5232
+    },
+    {
+      "epoch": 42.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1852,
+      "step": 5233
+    },
+    {
+      "epoch": 42.9,
+      "learning_rate": 3e-05,
+      "loss": 0.0895,
+      "step": 5234
+    },
+    {
+      "epoch": 42.91,
+      "learning_rate": 3e-05,
+      "loss": 0.1157,
+      "step": 5235
+    },
+    {
+      "epoch": 42.92,
+      "learning_rate": 3e-05,
+      "loss": 0.1091,
+      "step": 5236
+    },
+    {
+      "epoch": 42.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1235,
+      "step": 5237
+    },
+    {
+      "epoch": 42.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1631,
+      "step": 5238
+    },
+    {
+      "epoch": 42.94,
+      "learning_rate": 3e-05,
+      "loss": 0.1726,
+      "step": 5239
+    },
+    {
+      "epoch": 42.95,
+      "learning_rate": 3e-05,
+      "loss": 0.1442,
+      "step": 5240
+    },
+    {
+      "epoch": 42.96,
+      "learning_rate": 3e-05,
+      "loss": 0.1595,
+      "step": 5241
+    },
+    {
+      "epoch": 42.97,
+      "learning_rate": 3e-05,
+      "loss": 0.2695,
+      "step": 5242
+    },
+    {
+      "epoch": 42.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1324,
+      "step": 5243
+    },
+    {
+      "epoch": 42.98,
+      "learning_rate": 3e-05,
+      "loss": 0.2397,
+      "step": 5244
+    },
+    {
+      "epoch": 42.99,
+      "learning_rate": 3e-05,
+      "loss": 0.1231,
+      "step": 5245
+    },
+    {
+      "epoch": 43.0,
+      "learning_rate": 3e-05,
+      "loss": 0.21,
+      "step": 5246
+    },
+    {
+      "epoch": 43.0,
+      "eval_loss": 0.25830078125,
+      "eval_mae": 0.36669921875,
+      "eval_mse": 0.25830078125,
+      "eval_runtime": 51.1103,
+      "eval_samples_per_second": 3235.999,
+      "eval_steps_per_second": 0.274,
+      "step": 5246
+    },
+    {
+      "epoch": 43.01,
+      "learning_rate": 3e-05,
+      "loss": 0.2028,
+      "step": 5247
+    },
+    {
+      "epoch": 43.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1106,
+      "step": 5248
+    },
+    {
+      "epoch": 43.02,
+      "learning_rate": 3e-05,
+      "loss": 0.2062,
+      "step": 5249
+    },
+    {
+      "epoch": 43.03,
+      "learning_rate": 3e-05,
+      "loss": 0.0877,
+      "step": 5250
+    },
+    {
+      "epoch": 43.04,
+      "learning_rate": 3e-05,
+      "loss": 0.1632,
+      "step": 5251
+    },
+    {
+      "epoch": 43.05,
+      "learning_rate": 3e-05,
+      "loss": 0.0991,
+      "step": 5252
+    },
+    {
+      "epoch": 43.06,
+      "learning_rate": 3e-05,
+      "loss": 0.1718,
+      "step": 5253
+    },
+    {
+      "epoch": 43.07,
+      "learning_rate": 3e-05,
+      "loss": 0.176,
+      "step": 5254
+    },
+    {
+      "epoch": 43.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1694,
+      "step": 5255
+    },
+    {
+      "epoch": 43.08,
+      "learning_rate": 3e-05,
+      "loss": 0.1188,
+      "step": 5256
+    },
+    {
+      "epoch": 43.09,
+      "learning_rate": 3e-05,
+      "loss": 0.0986,
+      "step": 5257
+    },
+    {
+      "epoch": 43.1,
+      "learning_rate": 3e-05,
+      "loss": 0.1231,
+      "step": 5258
+    },
+    {
+      "epoch": 43.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1182,
+      "step": 5259
+    },
+    {
+      "epoch": 43.11,
+      "learning_rate": 3e-05,
+      "loss": 0.2007,
+      "step": 5260
+    },
+    {
+      "epoch": 43.12,
+      "learning_rate": 3e-05,
+      "loss": 0.1482,
+      "step": 5261
+    },
+    {
+      "epoch": 43.13,
+      "learning_rate": 3e-05,
+      "loss": 0.1421,
+      "step": 5262
+    },
+    {
+      "epoch": 43.14,
+      "learning_rate": 3e-05,
+      "loss": 0.1199,
+      "step": 5263
+    },
+    {
+      "epoch": 43.15,
+      "learning_rate": 3e-05,
+      "loss": 0.1208,
+      "step": 5264
+    },
+    {
+      "epoch": 43.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1676,
+      "step": 5265
+    },
+    {
+      "epoch": 43.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1976,
+      "step": 5266
+    },
+    {
+      "epoch": 43.17,
+      "learning_rate": 3e-05,
+      "loss": 0.1549,
+      "step": 5267
+    },
+    {
+      "epoch": 43.18,
+      "learning_rate": 3e-05,
+      "loss": 0.1918,
+      "step": 5268
+    },
+    {
+      "epoch": 43.19,
+      "learning_rate": 3e-05,
+      "loss": 0.2812,
+      "step": 5269
+    },
+    {
+      "epoch": 43.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1383,
+      "step": 5270
+    },
+    {
+      "epoch": 43.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1178,
+      "step": 5271
+    },
+    {
+      "epoch": 43.21,
+      "learning_rate": 3e-05,
+      "loss": 0.1418,
+      "step": 5272
+    },
+    {
+      "epoch": 43.22,
+      "learning_rate": 3e-05,
+      "loss": 0.1409,
+      "step": 5273
+    },
+    {
+      "epoch": 43.23,
+      "learning_rate": 3e-05,
+      "loss": 0.131,
+      "step": 5274
+    },
+    {
+      "epoch": 43.24,
+      "learning_rate": 3e-05,
+      "loss": 0.1458,
+      "step": 5275
+    },
+    {
+      "epoch": 43.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1124,
+      "step": 5276
+    },
+    {
+      "epoch": 43.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1292,
+      "step": 5277
+    },
+    {
+      "epoch": 43.26,
+      "learning_rate": 3e-05,
+      "loss": 0.1705,
+      "step": 5278
+    },
+    {
+      "epoch": 43.27,
+      "learning_rate": 3e-05,
+      "loss": 0.0908,
+      "step": 5279
+    },
+    {
+      "epoch": 43.28,
+      "learning_rate": 3e-05,
+      "loss": 0.1243,
+      "step": 5280
+    },
+    {
+      "epoch": 43.29,
+      "learning_rate": 3e-05,
+      "loss": 0.1061,
+      "step": 5281
+    },
+    {
+      "epoch": 43.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1497,
+      "step": 5282
+    },
+    {
+      "epoch": 43.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1785,
+      "step": 5283
+    },
+    {
+      "epoch": 43.31,
+      "learning_rate": 3e-05,
+      "loss": 0.1724,
+      "step": 5284
+    },
+    {
+      "epoch": 43.32,
+      "learning_rate": 3e-05,
+      "loss": 0.1595,
+      "step": 5285
+    },
+    {
+      "epoch": 43.33,
+      "learning_rate": 3e-05,
+      "loss": 0.1033,
+      "step": 5286
+    },
+    {
+      "epoch": 43.34,
+      "learning_rate": 3e-05,
+      "loss": 0.0902,
+      "step": 5287
+    },
+    {
+      "epoch": 43.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1558,
+      "step": 5288
+    },
+    {
+      "epoch": 43.35,
+      "learning_rate": 3e-05,
+      "loss": 0.0917,
+      "step": 5289
+    },
+    {
+      "epoch": 43.36,
+      "learning_rate": 3e-05,
+      "loss": 0.1342,
+      "step": 5290
+    },
+    {
+      "epoch": 43.37,
+      "learning_rate": 3e-05,
+      "loss": 0.1339,
+      "step": 5291
+    },
+    {
+      "epoch": 43.38,
+      "learning_rate": 3e-05,
+      "loss": 0.1311,
+      "step": 5292
+    },
+    {
+      "epoch": 43.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1129,
+      "step": 5293
+    },
+    {
+      "epoch": 43.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1263,
+      "step": 5294
+    },
+    {
+      "epoch": 43.4,
+      "learning_rate": 3e-05,
+      "loss": 0.13,
+      "step": 5295
+    },
+    {
+      "epoch": 43.41,
+      "learning_rate": 3e-05,
+      "loss": 0.146,
+      "step": 5296
+    },
+    {
+      "epoch": 43.42,
+      "learning_rate": 3e-05,
+      "loss": 0.0949,
+      "step": 5297
+    },
+    {
+      "epoch": 43.43,
+      "learning_rate": 3e-05,
+      "loss": 0.0821,
+      "step": 5298
+    },
+    {
+      "epoch": 43.43,
+      "learning_rate": 3e-05,
+      "loss": 0.1052,
+      "step": 5299
+    },
+    {
+      "epoch": 43.44,
+      "learning_rate": 3e-05,
+      "loss": 0.1027,
+      "step": 5300
+    },
+    {
+      "epoch": 43.45,
+      "learning_rate": 3e-05,
+      "loss": 0.1957,
+      "step": 5301
+    },
+    {
+      "epoch": 43.46,
+      "learning_rate": 3e-05,
+      "loss": 0.0864,
+      "step": 5302
+    },
+    {
+      "epoch": 43.47,
+      "learning_rate": 3e-05,
+      "loss": 0.1011,
+      "step": 5303
+    },
+    {
+      "epoch": 43.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1382,
+      "step": 5304
+    },
+    {
+      "epoch": 43.48,
+      "learning_rate": 3e-05,
+      "loss": 0.2252,
+      "step": 5305
+    },
+    {
+      "epoch": 43.49,
+      "learning_rate": 3e-05,
+      "loss": 0.1536,
+      "step": 5306
+    },
+    {
+      "epoch": 43.5,
+      "learning_rate": 3e-05,
+      "loss": 0.1606,
+      "step": 5307
+    },
+    {
+      "epoch": 43.51,
+      "learning_rate": 3e-05,
+      "loss": 0.1647,
+      "step": 5308
+    },
+    {
+      "epoch": 43.52,
+      "learning_rate": 3e-05,
+      "loss": 0.2019,
+      "step": 5309
+    },
+    {
+      "epoch": 43.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1356,
+      "step": 5310
+    },
+    {
+      "epoch": 43.53,
+      "learning_rate": 3e-05,
+      "loss": 0.1135,
+      "step": 5311
+    },
+    {
+      "epoch": 43.54,
+      "learning_rate": 3e-05,
+      "loss": 0.1444,
+      "step": 5312
+    },
+    {
+      "epoch": 43.55,
+      "learning_rate": 3e-05,
+      "loss": 0.1232,
+      "step": 5313
+    },
+    {
+      "epoch": 43.56,
+      "learning_rate": 3e-05,
+      "loss": 0.1063,
+      "step": 5314
+    },
+    {
+      "epoch": 43.57,
+      "learning_rate": 3e-05,
+      "loss": 0.1738,
+      "step": 5315
+    },
+    {
+      "epoch": 43.57,
+      "learning_rate": 3e-05,
+      "loss": 0.1125,
+      "step": 5316
+    },
+    {
+      "epoch": 43.58,
+      "learning_rate": 3e-05,
+      "loss": 0.1302,
+      "step": 5317
+    },
+    {
+      "epoch": 43.59,
+      "learning_rate": 3e-05,
+      "loss": 0.1254,
+      "step": 5318
+    },
+    {
+      "epoch": 43.6,
+      "learning_rate": 3e-05,
+      "loss": 0.1497,
+      "step": 5319
+    },
+    {
+      "epoch": 43.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1792,
+      "step": 5320
+    },
+    {
+      "epoch": 43.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1307,
+      "step": 5321
+    },
+    {
+      "epoch": 43.62,
+      "learning_rate": 3e-05,
+      "loss": 0.172,
+      "step": 5322
+    },
+    {
+      "epoch": 43.63,
+      "learning_rate": 3e-05,
+      "loss": 0.1809,
+      "step": 5323
+    },
+    {
+      "epoch": 43.64,
+      "learning_rate": 3e-05,
+      "loss": 0.1181,
+      "step": 5324
+    },
+    {
+      "epoch": 43.65,
+      "learning_rate": 3e-05,
+      "loss": 0.2338,
+      "step": 5325
+    },
+    {
+      "epoch": 43.66,
+      "learning_rate": 3e-05,
+      "loss": 0.2281,
+      "step": 5326
+    },
+    {
+      "epoch": 43.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1255,
+      "step": 5327
+    },
+    {
+      "epoch": 43.67,
+      "learning_rate": 3e-05,
+      "loss": 0.1202,
+      "step": 5328
+    },
+    {
+      "epoch": 43.68,
+      "learning_rate": 3e-05,
+      "loss": 0.191,
+      "step": 5329
+    },
+    {
+      "epoch": 43.69,
+      "learning_rate": 3e-05,
+      "loss": 0.1902,
+      "step": 5330
+    },
+    {
+      "epoch": 43.7,
+      "learning_rate": 3e-05,
+      "loss": 0.3313,
+      "step": 5331
+    },
+    {
+      "epoch": 43.7,
+      "learning_rate": 3e-05,
+      "loss": 0.1675,
+      "step": 5332
+    },
+    {
+      "epoch": 43.71,
+      "learning_rate": 3e-05,
+      "loss": 0.0865,
+      "step": 5333
+    },
+    {
+      "epoch": 43.72,
+      "learning_rate": 3e-05,
+      "loss": 0.1047,
+      "step": 5334
+    },
+    {
+      "epoch": 43.73,
+      "learning_rate": 3e-05,
+      "loss": 0.1218,
+      "step": 5335
+    },
+    {
+      "epoch": 43.74,
+      "learning_rate": 3e-05,
+      "loss": 0.1381,
+      "step": 5336
+    },
+    {
+      "epoch": 43.75,
+      "learning_rate": 3e-05,
+      "loss": 0.1663,
+      "step": 5337
+    },
+    {
+      "epoch": 43.75,
+      "learning_rate": 3e-05,
+      "loss": 0.1565,
+      "step": 5338
+    },
+    {
+      "epoch": 43.76,
+      "learning_rate": 3e-05,
+      "loss": 0.1776,
+      "step": 5339
+    },
+    {
+      "epoch": 43.77,
+      "learning_rate": 3e-05,
+      "loss": 0.1981,
+      "step": 5340
+    },
+    {
+      "epoch": 43.78,
+      "learning_rate": 3e-05,
+      "loss": 0.1274,
+      "step": 5341
+    },
+    {
+      "epoch": 43.79,
+      "learning_rate": 3e-05,
+      "loss": 0.1375,
+      "step": 5342
+    },
+    {
+      "epoch": 43.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1509,
+      "step": 5343
+    },
+    {
+      "epoch": 43.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1855,
+      "step": 5344
+    },
+    {
+      "epoch": 43.81,
+      "learning_rate": 3e-05,
+      "loss": 0.1398,
+      "step": 5345
+    },
+    {
+      "epoch": 43.82,
+      "learning_rate": 3e-05,
+      "loss": 0.1187,
+      "step": 5346
+    },
+    {
+      "epoch": 43.83,
+      "learning_rate": 3e-05,
+      "loss": 0.1771,
+      "step": 5347
+    },
+    {
+      "epoch": 43.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1107,
+      "step": 5348
+    },
+    {
+      "epoch": 43.84,
+      "learning_rate": 3e-05,
+      "loss": 0.0987,
+      "step": 5349
+    },
+    {
+      "epoch": 43.85,
+      "learning_rate": 3e-05,
+      "loss": 0.1583,
+      "step": 5350
+    },
+    {
+      "epoch": 43.86,
+      "learning_rate": 3e-05,
+      "loss": 0.2041,
+      "step": 5351
+    },
+    {
+      "epoch": 43.87,
+      "learning_rate": 3e-05,
+      "loss": 0.1069,
+      "step": 5352
+    },
+    {
+      "epoch": 43.88,
+      "learning_rate": 3e-05,
+      "loss": 0.1332,
+      "step": 5353
+    },
+    {
+      "epoch": 43.89,
+      "learning_rate": 3e-05,
+      "loss": 0.0911,
+      "step": 5354
+    },
+    {
+      "epoch": 43.89,
+      "learning_rate": 3e-05,
+      "loss": 0.2358,
+      "step": 5355
+    },
+    {
+      "epoch": 43.9,
+      "learning_rate": 3e-05,
+      "loss": 0.2017,
+      "step": 5356
+    },
+    {
+      "epoch": 43.91,
+      "learning_rate": 3e-05,
+      "loss": 0.1884,
+      "step": 5357
+    },
+    {
+      "epoch": 43.92,
+      "learning_rate": 3e-05,
+      "loss": 0.1953,
+      "step": 5358
+    },
+    {
+      "epoch": 43.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1827,
+      "step": 5359
+    },
+    {
+      "epoch": 43.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1065,
+      "step": 5360
+    },
+    {
+      "epoch": 43.94,
+      "learning_rate": 3e-05,
+      "loss": 0.1031,
+      "step": 5361
+    },
+    {
+      "epoch": 43.95,
+      "learning_rate": 3e-05,
+      "loss": 0.1174,
+      "step": 5362
+    },
+    {
+      "epoch": 43.96,
+      "learning_rate": 3e-05,
+      "loss": 0.1115,
+      "step": 5363
+    },
+    {
+      "epoch": 43.97,
+      "learning_rate": 3e-05,
+      "loss": 0.092,
+      "step": 5364
+    },
+    {
+      "epoch": 43.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1353,
+      "step": 5365
+    },
+    {
+      "epoch": 43.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1829,
+      "step": 5366
+    },
+    {
+      "epoch": 43.99,
+      "learning_rate": 3e-05,
+      "loss": 0.1792,
+      "step": 5367
+    },
+    {
+      "epoch": 44.0,
+      "learning_rate": 3e-05,
+      "loss": 0.1971,
+      "step": 5368
+    },
+    {
+      "epoch": 44.0,
+      "eval_loss": 0.255859375,
+      "eval_mae": 0.3642578125,
+      "eval_mse": 0.255859375,
+      "eval_runtime": 51.1096,
+      "eval_samples_per_second": 3236.049,
+      "eval_steps_per_second": 0.274,
+      "step": 5368
+    },
+    {
+      "epoch": 44.01,
+      "learning_rate": 3e-05,
+      "loss": 0.2024,
+      "step": 5369
+    },
+    {
+      "epoch": 44.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1098,
+      "step": 5370
+    },
+    {
+      "epoch": 44.02,
+      "learning_rate": 3e-05,
+      "loss": 0.0826,
+      "step": 5371
+    },
+    {
+      "epoch": 44.03,
+      "learning_rate": 3e-05,
+      "loss": 0.1285,
+      "step": 5372
+    },
+    {
+      "epoch": 44.04,
+      "learning_rate": 3e-05,
+      "loss": 0.113,
+      "step": 5373
+    },
+    {
+      "epoch": 44.05,
+      "learning_rate": 3e-05,
+      "loss": 0.1215,
+      "step": 5374
+    },
+    {
+      "epoch": 44.06,
+      "learning_rate": 3e-05,
+      "loss": 0.13,
+      "step": 5375
+    },
+    {
+      "epoch": 44.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1823,
+      "step": 5376
+    },
+    {
+      "epoch": 44.07,
+      "learning_rate": 3e-05,
+      "loss": 0.1486,
+      "step": 5377
+    },
+    {
+      "epoch": 44.08,
+      "learning_rate": 3e-05,
+      "loss": 0.1121,
+      "step": 5378
+    },
+    {
+      "epoch": 44.09,
+      "learning_rate": 3e-05,
+      "loss": 0.2288,
+      "step": 5379
+    },
+    {
+      "epoch": 44.1,
+      "learning_rate": 3e-05,
+      "loss": 0.0968,
+      "step": 5380
+    },
+    {
+      "epoch": 44.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1448,
+      "step": 5381
+    },
+    {
+      "epoch": 44.11,
+      "learning_rate": 3e-05,
+      "loss": 0.1785,
+      "step": 5382
+    },
+    {
+      "epoch": 44.12,
+      "learning_rate": 3e-05,
+      "loss": 0.1287,
+      "step": 5383
+    },
+    {
+      "epoch": 44.13,
+      "learning_rate": 3e-05,
+      "loss": 0.081,
+      "step": 5384
+    },
+    {
+      "epoch": 44.14,
+      "learning_rate": 3e-05,
+      "loss": 0.0763,
+      "step": 5385
+    },
+    {
+      "epoch": 44.15,
+      "learning_rate": 3e-05,
+      "loss": 0.1577,
+      "step": 5386
+    },
+    {
+      "epoch": 44.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1587,
+      "step": 5387
+    },
+    {
+      "epoch": 44.16,
+      "learning_rate": 3e-05,
+      "loss": 0.1672,
+      "step": 5388
+    },
+    {
+      "epoch": 44.17,
+      "learning_rate": 3e-05,
+      "loss": 0.1614,
+      "step": 5389
+    },
+    {
+      "epoch": 44.18,
+      "learning_rate": 3e-05,
+      "loss": 0.1153,
+      "step": 5390
+    },
+    {
+      "epoch": 44.19,
+      "learning_rate": 3e-05,
+      "loss": 0.2256,
+      "step": 5391
+    },
+    {
+      "epoch": 44.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1515,
+      "step": 5392
+    },
+    {
+      "epoch": 44.2,
+      "learning_rate": 3e-05,
+      "loss": 0.1735,
+      "step": 5393
+    },
+    {
+      "epoch": 44.21,
+      "learning_rate": 3e-05,
+      "loss": 0.1185,
+      "step": 5394
+    },
+    {
+      "epoch": 44.22,
+      "learning_rate": 3e-05,
+      "loss": 0.1239,
+      "step": 5395
+    },
+    {
+      "epoch": 44.23,
+      "learning_rate": 3e-05,
+      "loss": 0.173,
+      "step": 5396
+    },
+    {
+      "epoch": 44.24,
+      "learning_rate": 3e-05,
+      "loss": 0.075,
+      "step": 5397
+    },
+    {
+      "epoch": 44.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1445,
+      "step": 5398
+    },
+    {
+      "epoch": 44.25,
+      "learning_rate": 3e-05,
+      "loss": 0.1527,
+      "step": 5399
+    },
+    {
+      "epoch": 44.26,
+      "learning_rate": 3e-05,
+      "loss": 0.1229,
+      "step": 5400
+    },
+    {
+      "epoch": 44.27,
+      "learning_rate": 3e-05,
+      "loss": 0.1078,
+      "step": 5401
+    },
+    {
+      "epoch": 44.28,
+      "learning_rate": 3e-05,
+      "loss": 0.1672,
+      "step": 5402
+    },
+    {
+      "epoch": 44.29,
+      "learning_rate": 3e-05,
+      "loss": 0.2776,
+      "step": 5403
+    },
+    {
+      "epoch": 44.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1154,
+      "step": 5404
+    },
+    {
+      "epoch": 44.3,
+      "learning_rate": 3e-05,
+      "loss": 0.1285,
+      "step": 5405
+    },
+    {
+      "epoch": 44.31,
+      "learning_rate": 3e-05,
+      "loss": 0.1537,
+      "step": 5406
+    },
+    {
+      "epoch": 44.32,
+      "learning_rate": 3e-05,
+      "loss": 0.0809,
+      "step": 5407
+    },
+    {
+      "epoch": 44.33,
+      "learning_rate": 3e-05,
+      "loss": 0.1403,
+      "step": 5408
+    },
+    {
+      "epoch": 44.34,
+      "learning_rate": 3e-05,
+      "loss": 0.1221,
+      "step": 5409
+    },
+    {
+      "epoch": 44.34,
+      "learning_rate": 3e-05,
+      "loss": 0.0898,
+      "step": 5410
+    },
+    {
+      "epoch": 44.35,
+      "learning_rate": 3e-05,
+      "loss": 0.0854,
+      "step": 5411
+    },
+    {
+      "epoch": 44.36,
+      "learning_rate": 3e-05,
+      "loss": 0.0958,
+      "step": 5412
+    },
+    {
+      "epoch": 44.37,
+      "learning_rate": 3e-05,
+      "loss": 0.1989,
+      "step": 5413
+    },
+    {
+      "epoch": 44.38,
+      "learning_rate": 3e-05,
+      "loss": 0.1503,
+      "step": 5414
+    },
+    {
+      "epoch": 44.39,
+      "learning_rate": 3e-05,
+      "loss": 0.1711,
+      "step": 5415
+    },
+    {
+      "epoch": 44.39,
+      "learning_rate": 3e-05,
+      "loss": 0.0997,
+      "step": 5416
+    },
+    {
+      "epoch": 44.4,
+      "learning_rate": 3e-05,
+      "loss": 0.2303,
+      "step": 5417
+    },
+    {
+      "epoch": 44.41,
+      "learning_rate": 3e-05,
+      "loss": 0.1821,
+      "step": 5418
+    },
+    {
+      "epoch": 44.42,
+      "learning_rate": 3e-05,
+      "loss": 0.1993,
+      "step": 5419
+    },
+    {
+      "epoch": 44.43,
+      "learning_rate": 3e-05,
+      "loss": 0.2264,
+      "step": 5420
+    },
+    {
+      "epoch": 44.43,
+      "learning_rate": 3e-05,
+      "loss": 0.1636,
+      "step": 5421
+    },
+    {
+      "epoch": 44.44,
+      "learning_rate": 3e-05,
+      "loss": 0.2008,
+      "step": 5422
+    },
+    {
+      "epoch": 44.45,
+      "learning_rate": 3e-05,
+      "loss": 0.119,
+      "step": 5423
+    },
+    {
+      "epoch": 44.46,
+      "learning_rate": 3e-05,
+      "loss": 0.1786,
+      "step": 5424
+    },
+    {
+      "epoch": 44.47,
+      "learning_rate": 3e-05,
+      "loss": 0.1862,
+      "step": 5425
+    },
+    {
+      "epoch": 44.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1348,
+      "step": 5426
+    },
+    {
+      "epoch": 44.48,
+      "learning_rate": 3e-05,
+      "loss": 0.1317,
+      "step": 5427
+    },
+    {
+      "epoch": 44.49,
+      "learning_rate": 3e-05,
+      "loss": 0.1971,
+      "step": 5428
+    },
+    {
+      "epoch": 44.5,
+      "learning_rate": 3e-05,
+      "loss": 0.1002,
+      "step": 5429
+    },
+    {
+      "epoch": 44.51,
+      "learning_rate": 3e-05,
+      "loss": 0.1334,
+      "step": 5430
+    },
+    {
+      "epoch": 44.52,
+      "learning_rate": 3e-05,
+      "loss": 0.1613,
+      "step": 5431
+    },
+    {
+      "epoch": 44.52,
+      "learning_rate": 3e-05,
+      "loss": 0.2253,
+      "step": 5432
+    },
+    {
+      "epoch": 44.53,
+      "learning_rate": 3e-05,
+      "loss": 0.0932,
+      "step": 5433
+    },
+    {
+      "epoch": 44.54,
+      "learning_rate": 3e-05,
+      "loss": 0.1373,
+      "step": 5434
+    },
+    {
+      "epoch": 44.55,
+      "learning_rate": 3e-05,
+      "loss": 0.1158,
+      "step": 5435
+    },
+    {
+      "epoch": 44.56,
+      "learning_rate": 3e-05,
+      "loss": 0.1874,
+      "step": 5436
+    },
+    {
+      "epoch": 44.57,
+      "learning_rate": 3e-05,
+      "loss": 0.1368,
+      "step": 5437
+    },
+    {
+      "epoch": 44.57,
+      "learning_rate": 3e-05,
+      "loss": 0.1598,
+      "step": 5438
+    },
+    {
+      "epoch": 44.58,
+      "learning_rate": 3e-05,
+      "loss": 0.1798,
+      "step": 5439
+    },
+    {
+      "epoch": 44.59,
+      "learning_rate": 3e-05,
+      "loss": 0.1167,
+      "step": 5440
+    },
+    {
+      "epoch": 44.6,
+      "learning_rate": 3e-05,
+      "loss": 0.1123,
+      "step": 5441
+    },
+    {
+      "epoch": 44.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1233,
+      "step": 5442
+    },
+    {
+      "epoch": 44.61,
+      "learning_rate": 3e-05,
+      "loss": 0.1964,
+      "step": 5443
+    },
+    {
+      "epoch": 44.62,
+      "learning_rate": 3e-05,
+      "loss": 0.152,
+      "step": 5444
+    },
+    {
+      "epoch": 44.63,
+      "learning_rate": 3e-05,
+      "loss": 0.2015,
+      "step": 5445
+    },
+    {
+      "epoch": 44.64,
+      "learning_rate": 3e-05,
+      "loss": 0.1396,
+      "step": 5446
+    },
+    {
+      "epoch": 44.65,
+      "learning_rate": 3e-05,
+      "loss": 0.1021,
+      "step": 5447
+    },
+    {
+      "epoch": 44.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1733,
+      "step": 5448
+    },
+    {
+      "epoch": 44.66,
+      "learning_rate": 3e-05,
+      "loss": 0.1743,
+      "step": 5449
+    },
+    {
+      "epoch": 44.67,
+      "learning_rate": 3e-05,
+      "loss": 0.0993,
+      "step": 5450
+    },
+    {
+      "epoch": 44.68,
+      "learning_rate": 3e-05,
+      "loss": 0.1552,
+      "step": 5451
+    },
+    {
+      "epoch": 44.69,
+      "learning_rate": 3e-05,
+      "loss": 0.1392,
+      "step": 5452
+    },
+    {
+      "epoch": 44.7,
+      "learning_rate": 3e-05,
+      "loss": 0.2113,
+      "step": 5453
+    },
+    {
+      "epoch": 44.7,
+      "learning_rate": 3e-05,
+      "loss": 0.0817,
+      "step": 5454
+    },
+    {
+      "epoch": 44.71,
+      "learning_rate": 3e-05,
+      "loss": 0.1437,
+      "step": 5455
+    },
+    {
+      "epoch": 44.72,
+      "learning_rate": 3e-05,
+      "loss": 0.1732,
+      "step": 5456
+    },
+    {
+      "epoch": 44.73,
+      "learning_rate": 3e-05,
+      "loss": 0.147,
+      "step": 5457
+    },
+    {
+      "epoch": 44.74,
+      "learning_rate": 3e-05,
+      "loss": 0.147,
+      "step": 5458
+    },
+    {
+      "epoch": 44.75,
+      "learning_rate": 3e-05,
+      "loss": 0.1416,
+      "step": 5459
+    },
+    {
+      "epoch": 44.75,
+      "learning_rate": 3e-05,
+      "loss": 0.1129,
+      "step": 5460
+    },
+    {
+      "epoch": 44.76,
+      "learning_rate": 3e-05,
+      "loss": 0.1588,
+      "step": 5461
+    },
+    {
+      "epoch": 44.77,
+      "learning_rate": 3e-05,
+      "loss": 0.102,
+      "step": 5462
+    },
+    {
+      "epoch": 44.78,
+      "learning_rate": 3e-05,
+      "loss": 0.1242,
+      "step": 5463
+    },
+    {
+      "epoch": 44.79,
+      "learning_rate": 3e-05,
+      "loss": 0.0728,
+      "step": 5464
+    },
+    {
+      "epoch": 44.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1648,
+      "step": 5465
+    },
+    {
+      "epoch": 44.8,
+      "learning_rate": 3e-05,
+      "loss": 0.1267,
+      "step": 5466
+    },
+    {
+      "epoch": 44.81,
+      "learning_rate": 3e-05,
+      "loss": 0.1007,
+      "step": 5467
+    },
+    {
+      "epoch": 44.82,
+      "learning_rate": 3e-05,
+      "loss": 0.0762,
+      "step": 5468
+    },
+    {
+      "epoch": 44.83,
+      "learning_rate": 3e-05,
+      "loss": 0.246,
+      "step": 5469
+    },
+    {
+      "epoch": 44.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1392,
+      "step": 5470
+    },
+    {
+      "epoch": 44.84,
+      "learning_rate": 3e-05,
+      "loss": 0.1984,
+      "step": 5471
+    },
+    {
+      "epoch": 44.85,
+      "learning_rate": 3e-05,
+      "loss": 0.222,
+      "step": 5472
+    },
+    {
+      "epoch": 44.86,
+      "learning_rate": 3e-05,
+      "loss": 0.1438,
+      "step": 5473
+    },
+    {
+      "epoch": 44.87,
+      "learning_rate": 3e-05,
+      "loss": 0.2061,
+      "step": 5474
+    },
+    {
+      "epoch": 44.88,
+      "learning_rate": 3e-05,
+      "loss": 0.1004,
+      "step": 5475
+    },
+    {
+      "epoch": 44.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1251,
+      "step": 5476
+    },
+    {
+      "epoch": 44.89,
+      "learning_rate": 3e-05,
+      "loss": 0.1393,
+      "step": 5477
+    },
+    {
+      "epoch": 44.9,
+      "learning_rate": 3e-05,
+      "loss": 0.2266,
+      "step": 5478
+    },
+    {
+      "epoch": 44.91,
+      "learning_rate": 3e-05,
+      "loss": 0.1616,
+      "step": 5479
+    },
+    {
+      "epoch": 44.92,
+      "learning_rate": 3e-05,
+      "loss": 0.149,
+      "step": 5480
+    },
+    {
+      "epoch": 44.93,
+      "learning_rate": 3e-05,
+      "loss": 0.2231,
+      "step": 5481
+    },
+    {
+      "epoch": 44.93,
+      "learning_rate": 3e-05,
+      "loss": 0.1266,
+      "step": 5482
+    },
+    {
+      "epoch": 44.94,
+      "learning_rate": 3e-05,
+      "loss": 0.1578,
+      "step": 5483
+    },
+    {
+      "epoch": 44.95,
+      "learning_rate": 3e-05,
+      "loss": 0.1176,
+      "step": 5484
+    },
+    {
+      "epoch": 44.96,
+      "learning_rate": 3e-05,
+      "loss": 0.0914,
+      "step": 5485
+    },
+    {
+      "epoch": 44.97,
+      "learning_rate": 3e-05,
+      "loss": 0.2427,
+      "step": 5486
+    },
+    {
+      "epoch": 44.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1057,
+      "step": 5487
+    },
+    {
+      "epoch": 44.98,
+      "learning_rate": 3e-05,
+      "loss": 0.1436,
+      "step": 5488
+    },
+    {
+      "epoch": 44.99,
+      "learning_rate": 3e-05,
+      "loss": 0.1412,
+      "step": 5489
+    },
+    {
+      "epoch": 45.0,
+      "learning_rate": 3e-05,
+      "loss": 0.0724,
+      "step": 5490
+    },
+    {
+      "epoch": 45.0,
+      "eval_loss": 0.26123046875,
+      "eval_mae": 0.368896484375,
+      "eval_mse": 0.26123046875,
+      "eval_runtime": 51.1146,
+      "eval_samples_per_second": 3235.731,
+      "eval_steps_per_second": 0.274,
+      "step": 5490
+    },
+    {
+      "epoch": 45.0,
+      "step": 5490,
+      "total_flos": 0.0,
+      "train_loss": 0.25706558644445865,
+      "train_runtime": 38020.7,
+      "train_samples_per_second": 1761.781,
+      "train_steps_per_second": 0.144
+    }
+  ],
+  "max_steps": 5490,
+  "num_train_epochs": 45,
+  "total_flos": 0.0,
+  "trial_name": null,
+  "trial_params": null
+}