diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,24181 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.8109627547434997,
+  "eval_steps": 200,
+  "global_step": 4000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-08,
+      "loss": 1.8695,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-08,
+      "loss": 1.8539,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-08,
+      "loss": 1.8445,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-08,
+      "loss": 1.8934,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-08,
+      "loss": 1.8662,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-08,
+      "loss": 1.8703,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7e-08,
+      "loss": 1.8473,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8e-08,
+      "loss": 1.8929,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9e-08,
+      "loss": 1.8763,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1e-07,
+      "loss": 1.8732,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0999999999999999e-07,
+      "loss": 1.8554,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2e-07,
+      "loss": 1.8738,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3e-07,
+      "loss": 1.8404,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4e-07,
+      "loss": 1.8802,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5e-07,
+      "loss": 1.8551,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6e-07,
+      "loss": 1.8597,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7000000000000001e-07,
+      "loss": 1.8663,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8e-07,
+      "loss": 1.8944,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8999999999999998e-07,
+      "loss": 1.8801,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-07,
+      "loss": 1.8502,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0999999999999997e-07,
+      "loss": 1.8727,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1999999999999998e-07,
+      "loss": 1.8872,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3e-07,
+      "loss": 1.8561,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4e-07,
+      "loss": 1.8488,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5e-07,
+      "loss": 1.8277,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6e-07,
+      "loss": 1.8605,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7e-07,
+      "loss": 1.8441,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8e-07,
+      "loss": 1.8126,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9e-07,
+      "loss": 1.8737,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3e-07,
+      "loss": 1.8349,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1e-07,
+      "loss": 1.8518,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2e-07,
+      "loss": 1.8745,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3e-07,
+      "loss": 1.8617,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4000000000000003e-07,
+      "loss": 1.8391,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5e-07,
+      "loss": 1.8687,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6e-07,
+      "loss": 1.859,
+      "step": 36
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7e-07,
+      "loss": 1.8731,
+      "step": 37
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7999999999999996e-07,
+      "loss": 1.8434,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8999999999999997e-07,
+      "loss": 1.8483,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4e-07,
+      "loss": 1.8452,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.0999999999999994e-07,
+      "loss": 1.8264,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.1999999999999995e-07,
+      "loss": 1.8339,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.2999999999999996e-07,
+      "loss": 1.778,
+      "step": 43
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3999999999999997e-07,
+      "loss": 1.825,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.5e-07,
+      "loss": 1.8307,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.6e-07,
+      "loss": 1.8182,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.6999999999999995e-07,
+      "loss": 1.8343,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8e-07,
+      "loss": 1.8334,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9e-07,
+      "loss": 1.8553,
+      "step": 49
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5e-07,
+      "loss": 1.8333,
+      "step": 50
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.1e-07,
+      "loss": 1.84,
+      "step": 51
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.2e-07,
+      "loss": 1.8163,
+      "step": 52
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.3e-07,
+      "loss": 1.8197,
+      "step": 53
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.4e-07,
+      "loss": 1.828,
+      "step": 54
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5e-07,
+      "loss": 1.8322,
+      "step": 55
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6e-07,
+      "loss": 1.8229,
+      "step": 56
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.699999999999999e-07,
+      "loss": 1.836,
+      "step": 57
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8e-07,
+      "loss": 1.7938,
+      "step": 58
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9e-07,
+      "loss": 1.8197,
+      "step": 59
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6e-07,
+      "loss": 1.8139,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.1e-07,
+      "loss": 1.8148,
+      "step": 61
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.2e-07,
+      "loss": 1.8255,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.3e-07,
+      "loss": 1.8267,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.4e-07,
+      "loss": 1.8394,
+      "step": 64
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.5e-07,
+      "loss": 1.8242,
+      "step": 65
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.6e-07,
+      "loss": 1.7853,
+      "step": 66
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.7e-07,
+      "loss": 1.8293,
+      "step": 67
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 1.7911,
+      "step": 68
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.9e-07,
+      "loss": 1.7815,
+      "step": 69
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7e-07,
+      "loss": 1.8137,
+      "step": 70
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.1e-07,
+      "loss": 1.8144,
+      "step": 71
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.2e-07,
+      "loss": 1.7956,
+      "step": 72
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.3e-07,
+      "loss": 1.8274,
+      "step": 73
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.4e-07,
+      "loss": 1.7906,
+      "step": 74
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.5e-07,
+      "loss": 1.7972,
+      "step": 75
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.599999999999999e-07,
+      "loss": 1.8101,
+      "step": 76
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.699999999999999e-07,
+      "loss": 1.8078,
+      "step": 77
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.799999999999999e-07,
+      "loss": 1.819,
+      "step": 78
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.9e-07,
+      "loss": 1.8017,
+      "step": 79
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8e-07,
+      "loss": 1.7866,
+      "step": 80
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.1e-07,
+      "loss": 1.7912,
+      "step": 81
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.199999999999999e-07,
+      "loss": 1.7933,
+      "step": 82
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.299999999999999e-07,
+      "loss": 1.7805,
+      "step": 83
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.399999999999999e-07,
+      "loss": 1.8109,
+      "step": 84
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.499999999999999e-07,
+      "loss": 1.8228,
+      "step": 85
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.599999999999999e-07,
+      "loss": 1.8178,
+      "step": 86
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.699999999999999e-07,
+      "loss": 1.7763,
+      "step": 87
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.799999999999999e-07,
+      "loss": 1.7953,
+      "step": 88
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.9e-07,
+      "loss": 1.782,
+      "step": 89
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9e-07,
+      "loss": 1.7932,
+      "step": 90
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.1e-07,
+      "loss": 1.7644,
+      "step": 91
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.2e-07,
+      "loss": 1.8147,
+      "step": 92
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.3e-07,
+      "loss": 1.7743,
+      "step": 93
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.399999999999999e-07,
+      "loss": 1.7866,
+      "step": 94
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.499999999999999e-07,
+      "loss": 1.7591,
+      "step": 95
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.6e-07,
+      "loss": 1.7776,
+      "step": 96
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.7e-07,
+      "loss": 1.7788,
+      "step": 97
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.8e-07,
+      "loss": 1.8216,
+      "step": 98
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9e-07,
+      "loss": 1.7772,
+      "step": 99
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1e-06,
+      "loss": 1.7798,
+      "step": 100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.0099999999999999e-06,
+      "loss": 1.8094,
+      "step": 101
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.02e-06,
+      "loss": 1.7753,
+      "step": 102
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.0299999999999999e-06,
+      "loss": 1.7982,
+      "step": 103
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.04e-06,
+      "loss": 1.7767,
+      "step": 104
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.05e-06,
+      "loss": 1.7694,
+      "step": 105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.06e-06,
+      "loss": 1.7721,
+      "step": 106
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.07e-06,
+      "loss": 1.7611,
+      "step": 107
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.08e-06,
+      "loss": 1.772,
+      "step": 108
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.09e-06,
+      "loss": 1.8015,
+      "step": 109
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1e-06,
+      "loss": 1.801,
+      "step": 110
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.11e-06,
+      "loss": 1.7951,
+      "step": 111
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.12e-06,
+      "loss": 1.7986,
+      "step": 112
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1299999999999998e-06,
+      "loss": 1.8062,
+      "step": 113
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1399999999999999e-06,
+      "loss": 1.7744,
+      "step": 114
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1499999999999998e-06,
+      "loss": 1.7799,
+      "step": 115
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.16e-06,
+      "loss": 1.787,
+      "step": 116
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1699999999999998e-06,
+      "loss": 1.7957,
+      "step": 117
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.18e-06,
+      "loss": 1.7931,
+      "step": 118
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1899999999999998e-06,
+      "loss": 1.8035,
+      "step": 119
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.2e-06,
+      "loss": 1.7578,
+      "step": 120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2099999999999998e-06,
+      "loss": 1.7859,
+      "step": 121
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.22e-06,
+      "loss": 1.785,
+      "step": 122
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2299999999999999e-06,
+      "loss": 1.7633,
+      "step": 123
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.24e-06,
+      "loss": 1.7513,
+      "step": 124
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2499999999999999e-06,
+      "loss": 1.7618,
+      "step": 125
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.26e-06,
+      "loss": 1.7817,
+      "step": 126
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.27e-06,
+      "loss": 1.7392,
+      "step": 127
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.28e-06,
+      "loss": 1.7678,
+      "step": 128
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.29e-06,
+      "loss": 1.7661,
+      "step": 129
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3e-06,
+      "loss": 1.7745,
+      "step": 130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.31e-06,
+      "loss": 1.7805,
+      "step": 131
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.32e-06,
+      "loss": 1.7743,
+      "step": 132
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.33e-06,
+      "loss": 1.7652,
+      "step": 133
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.34e-06,
+      "loss": 1.7631,
+      "step": 134
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.35e-06,
+      "loss": 1.7678,
+      "step": 135
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3600000000000001e-06,
+      "loss": 1.7616,
+      "step": 136
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.37e-06,
+      "loss": 1.7246,
+      "step": 137
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.38e-06,
+      "loss": 1.7683,
+      "step": 138
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3899999999999998e-06,
+      "loss": 1.7862,
+      "step": 139
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4e-06,
+      "loss": 1.7939,
+      "step": 140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4099999999999998e-06,
+      "loss": 1.7763,
+      "step": 141
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.42e-06,
+      "loss": 1.7464,
+      "step": 142
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4299999999999999e-06,
+      "loss": 1.7775,
+      "step": 143
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.44e-06,
+      "loss": 1.7779,
+      "step": 144
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.4499999999999999e-06,
+      "loss": 1.7986,
+      "step": 145
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.46e-06,
+      "loss": 1.7745,
+      "step": 146
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.47e-06,
+      "loss": 1.7811,
+      "step": 147
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.48e-06,
+      "loss": 1.7789,
+      "step": 148
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.49e-06,
+      "loss": 1.7614,
+      "step": 149
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5e-06,
+      "loss": 1.7854,
+      "step": 150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.51e-06,
+      "loss": 1.7122,
+      "step": 151
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5199999999999998e-06,
+      "loss": 1.7446,
+      "step": 152
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.53e-06,
+      "loss": 1.7521,
+      "step": 153
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5399999999999999e-06,
+      "loss": 1.7478,
+      "step": 154
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.55e-06,
+      "loss": 1.773,
+      "step": 155
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5599999999999999e-06,
+      "loss": 1.7662,
+      "step": 156
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.57e-06,
+      "loss": 1.7309,
+      "step": 157
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.58e-06,
+      "loss": 1.7519,
+      "step": 158
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.59e-06,
+      "loss": 1.7853,
+      "step": 159
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.6e-06,
+      "loss": 1.7849,
+      "step": 160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.61e-06,
+      "loss": 1.7779,
+      "step": 161
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.62e-06,
+      "loss": 1.7741,
+      "step": 162
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.6299999999999999e-06,
+      "loss": 1.7461,
+      "step": 163
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6399999999999998e-06,
+      "loss": 1.7552,
+      "step": 164
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6499999999999999e-06,
+      "loss": 1.7706,
+      "step": 165
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6599999999999998e-06,
+      "loss": 1.7291,
+      "step": 166
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6699999999999999e-06,
+      "loss": 1.7707,
+      "step": 167
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6799999999999998e-06,
+      "loss": 1.7613,
+      "step": 168
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.69e-06,
+      "loss": 1.7625,
+      "step": 169
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6999999999999998e-06,
+      "loss": 1.7612,
+      "step": 170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.71e-06,
+      "loss": 1.7645,
+      "step": 171
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7199999999999998e-06,
+      "loss": 1.747,
+      "step": 172
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.73e-06,
+      "loss": 1.764,
+      "step": 173
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7399999999999999e-06,
+      "loss": 1.7707,
+      "step": 174
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.75e-06,
+      "loss": 1.757,
+      "step": 175
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7599999999999999e-06,
+      "loss": 1.7755,
+      "step": 176
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.77e-06,
+      "loss": 1.7731,
+      "step": 177
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.78e-06,
+      "loss": 1.7314,
+      "step": 178
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.79e-06,
+      "loss": 1.7639,
+      "step": 179
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8e-06,
+      "loss": 1.7528,
+      "step": 180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.81e-06,
+      "loss": 1.7731,
+      "step": 181
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.82e-06,
+      "loss": 1.7232,
+      "step": 182
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.83e-06,
+      "loss": 1.7652,
+      "step": 183
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.84e-06,
+      "loss": 1.7381,
+      "step": 184
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.85e-06,
+      "loss": 1.7685,
+      "step": 185
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.86e-06,
+      "loss": 1.7352,
+      "step": 186
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.87e-06,
+      "loss": 1.7711,
+      "step": 187
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8799999999999998e-06,
+      "loss": 1.7484,
+      "step": 188
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.89e-06,
+      "loss": 1.7522,
+      "step": 189
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8999999999999998e-06,
+      "loss": 1.7685,
+      "step": 190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.91e-06,
+      "loss": 1.761,
+      "step": 191
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.92e-06,
+      "loss": 1.7487,
+      "step": 192
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9299999999999997e-06,
+      "loss": 1.7566,
+      "step": 193
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94e-06,
+      "loss": 1.732,
+      "step": 194
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.95e-06,
+      "loss": 1.7697,
+      "step": 195
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.96e-06,
+      "loss": 1.7397,
+      "step": 196
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9699999999999998e-06,
+      "loss": 1.7614,
+      "step": 197
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.98e-06,
+      "loss": 1.747,
+      "step": 198
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99e-06,
+      "loss": 1.7719,
+      "step": 199
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-06,
+      "loss": 1.7632,
+      "step": 200
+    },
+    {
+      "epoch": 0.14,
+      "eval_loss": 1.747337818145752,
+      "eval_runtime": 51.65,
+      "eval_samples_per_second": 38.722,
+      "eval_steps_per_second": 0.097,
+      "step": 200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9999990671452865e-06,
+      "loss": 1.7213,
+      "step": 201
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9999962685828873e-06,
+      "loss": 1.7784,
+      "step": 202
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999991604318023e-06,
+      "loss": 1.7742,
+      "step": 203
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999985074359396e-06,
+      "loss": 1.748,
+      "step": 204
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9999766787191892e-06,
+      "loss": 1.7546,
+      "step": 205
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999966417413067e-06,
+      "loss": 1.7263,
+      "step": 206
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999954290460173e-06,
+      "loss": 1.7125,
+      "step": 207
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999402978831336e-06,
+      "loss": 1.7282,
+      "step": 208
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999244397080543e-06,
+      "loss": 1.7501,
+      "step": 209
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999906715964522e-06,
+      "loss": 1.7733,
+      "step": 210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999887126685604e-06,
+      "loss": 1.7288,
+      "step": 211
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999865671907848e-06,
+      "loss": 1.7599,
+      "step": 212
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999842351671283e-06,
+      "loss": 1.7397,
+      "step": 213
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999817166019417e-06,
+      "loss": 1.7445,
+      "step": 214
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99979011499924e-06,
+      "loss": 1.7534,
+      "step": 215
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99976119866122e-06,
+      "loss": 1.7944,
+      "step": 216
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999730417059308e-06,
+      "loss": 1.7349,
+      "step": 217
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999697770250933e-06,
+      "loss": 1.732,
+      "step": 218
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999663258297004e-06,
+      "loss": 1.7607,
+      "step": 219
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9996268812619105e-06,
+      "loss": 1.7396,
+      "step": 220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9995886392135216e-06,
+      "loss": 1.7433,
+      "step": 221
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9995485322231864e-06,
+      "loss": 1.7649,
+      "step": 222
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9995065603657316e-06,
+      "loss": 1.7522,
+      "step": 223
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999462723719465e-06,
+      "loss": 1.7759,
+      "step": 224
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999417022366174e-06,
+      "loss": 1.7445,
+      "step": 225
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9993694563911226e-06,
+      "loss": 1.7522,
+      "step": 226
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9993200258830566e-06,
+      "loss": 1.7503,
+      "step": 227
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9992687309341972e-06,
+      "loss": 1.7486,
+      "step": 228
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9992155716402473e-06,
+      "loss": 1.7713,
+      "step": 229
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9991605481003864e-06,
+      "loss": 1.764,
+      "step": 230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9991036604172723e-06,
+      "loss": 1.7504,
+      "step": 231
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99904490869704e-06,
+      "loss": 1.7478,
+      "step": 232
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.998984293049305e-06,
+      "loss": 1.752,
+      "step": 233
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9989218135871567e-06,
+      "loss": 1.7732,
+      "step": 234
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9988574704271648e-06,
+      "loss": 1.771,
+      "step": 235
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9987912636893743e-06,
+      "loss": 1.739,
+      "step": 236
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.998723193497308e-06,
+      "loss": 1.7195,
+      "step": 237
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.998653259977965e-06,
+      "loss": 1.7559,
+      "step": 238
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.998581463261821e-06,
+      "loss": 1.7423,
+      "step": 239
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.998507803482828e-06,
+      "loss": 1.7508,
+      "step": 240
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.998432280778413e-06,
+      "loss": 1.7404,
+      "step": 241
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9983548952894805e-06,
+      "loss": 1.7358,
+      "step": 242
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.998275647160409e-06,
+      "loss": 1.7371,
+      "step": 243
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9981945365390516e-06,
+      "loss": 1.7384,
+      "step": 244
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9981115635767376e-06,
+      "loss": 1.7273,
+      "step": 245
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9980267284282714e-06,
+      "loss": 1.7208,
+      "step": 246
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9979400312519296e-06,
+      "loss": 1.7479,
+      "step": 247
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9978514722094646e-06,
+      "loss": 1.7301,
+      "step": 248
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9977610514661015e-06,
+      "loss": 1.7349,
+      "step": 249
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9976687691905393e-06,
+      "loss": 1.7472,
+      "step": 250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9975746255549496e-06,
+      "loss": 1.7495,
+      "step": 251
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.997478620734977e-06,
+      "loss": 1.7144,
+      "step": 252
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9973807549097396e-06,
+      "loss": 1.7405,
+      "step": 253
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9972810282618252e-06,
+      "loss": 1.7223,
+      "step": 254
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.997179440977296e-06,
+      "loss": 1.7381,
+      "step": 255
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.997075993245683e-06,
+      "loss": 1.7527,
+      "step": 256
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9969706852599912e-06,
+      "loss": 1.7464,
+      "step": 257
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9968635172166935e-06,
+      "loss": 1.7462,
+      "step": 258
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.996754489315735e-06,
+      "loss": 1.7421,
+      "step": 259
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9966436017605294e-06,
+      "loss": 1.7274,
+      "step": 260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9965308547579613e-06,
+      "loss": 1.7389,
+      "step": 261
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9964162485183834e-06,
+      "loss": 1.7552,
+      "step": 262
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.996299783255618e-06,
+      "loss": 1.7527,
+      "step": 263
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9961814591869557e-06,
+      "loss": 1.749,
+      "step": 264
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.996061276533154e-06,
+      "loss": 1.6966,
+      "step": 265
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9959392355184387e-06,
+      "loss": 1.7536,
+      "step": 266
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9958153363705042e-06,
+      "loss": 1.7774,
+      "step": 267
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.995689579320509e-06,
+      "loss": 1.7308,
+      "step": 268
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99556196460308e-06,
+      "loss": 1.756,
+      "step": 269
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9954324924563086e-06,
+      "loss": 1.7146,
+      "step": 270
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.995301163121753e-06,
+      "loss": 1.7602,
+      "step": 271
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9951679768444345e-06,
+      "loss": 1.7253,
+      "step": 272
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.995032933872841e-06,
+      "loss": 1.7505,
+      "step": 273
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994896034458923e-06,
+      "loss": 1.7278,
+      "step": 274
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9947572788580945e-06,
+      "loss": 1.7489,
+      "step": 275
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994616667329234e-06,
+      "loss": 1.7535,
+      "step": 276
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944742001346818e-06,
+      "loss": 1.7214,
+      "step": 277
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9943298775402397e-06,
+      "loss": 1.7178,
+      "step": 278
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.994183699815172e-06,
+      "loss": 1.7263,
+      "step": 279
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9940356672322033e-06,
+      "loss": 1.7095,
+      "step": 280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9938857800675206e-06,
+      "loss": 1.7663,
+      "step": 281
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993734038600769e-06,
+      "loss": 1.7433,
+      "step": 282
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935804431150535e-06,
+      "loss": 1.7451,
+      "step": 283
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934249938969395e-06,
+      "loss": 1.7442,
+      "step": 284
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99326769123645e-06,
+      "loss": 1.7389,
+      "step": 285
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931085354270657e-06,
+      "loss": 1.7319,
+      "step": 286
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929475267657252e-06,
+      "loss": 1.7478,
+      "step": 287
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927846655528236e-06,
+      "loss": 1.7553,
+      "step": 288
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926199520922132e-06,
+      "loss": 1.7317,
+      "step": 289
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924533866912014e-06,
+      "loss": 1.7465,
+      "step": 290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9922849696605506e-06,
+      "loss": 1.7285,
+      "step": 291
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921147013144777e-06,
+      "loss": 1.7526,
+      "step": 292
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919425819706546e-06,
+      "loss": 1.7375,
+      "step": 293
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917686119502057e-06,
+      "loss": 1.7691,
+      "step": 294
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991592791577708e-06,
+      "loss": 1.7457,
+      "step": 295
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914151211811924e-06,
+      "loss": 1.7523,
+      "step": 296
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912356010921394e-06,
+      "loss": 1.7344,
+      "step": 297
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991054231645481e-06,
+      "loss": 1.7798,
+      "step": 298
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9908710131796007e-06,
+      "loss": 1.7463,
+      "step": 299
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9906859460363304e-06,
+      "loss": 1.6971,
+      "step": 300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.990499030560952e-06,
+      "loss": 1.6904,
+      "step": 301
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9903102671021953e-06,
+      "loss": 1.7162,
+      "step": 302
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9901196560122383e-06,
+      "loss": 1.7596,
+      "step": 303
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9899271976467055e-06,
+      "loss": 1.7355,
+      "step": 304
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.989732892364668e-06,
+      "loss": 1.7266,
+      "step": 305
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895367405286435e-06,
+      "loss": 1.7335,
+      "step": 306
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9893387425045948e-06,
+      "loss": 1.7421,
+      "step": 307
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9891388986619275e-06,
+      "loss": 1.7542,
+      "step": 308
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.988937209373493e-06,
+      "loss": 1.7399,
+      "step": 309
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9887336750155845e-06,
+      "loss": 1.7233,
+      "step": 310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.988528295967939e-06,
+      "loss": 1.6922,
+      "step": 311
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9883210726137324e-06,
+      "loss": 1.7243,
+      "step": 312
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9881120053395843e-06,
+      "loss": 1.7529,
+      "step": 313
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.987901094535553e-06,
+      "loss": 1.7332,
+      "step": 314
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9876883405951377e-06,
+      "loss": 1.7162,
+      "step": 315
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9874737439152748e-06,
+      "loss": 1.7421,
+      "step": 316
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.987257304896339e-06,
+      "loss": 1.7095,
+      "step": 317
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.987039023942143e-06,
+      "loss": 1.7447,
+      "step": 318
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.986818901459936e-06,
+      "loss": 1.7364,
+      "step": 319
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.986596937860402e-06,
+      "loss": 1.7539,
+      "step": 320
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9863731335576607e-06,
+      "loss": 1.7313,
+      "step": 321
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9861474889692664e-06,
+      "loss": 1.7442,
+      "step": 322
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9859200045162052e-06,
+      "loss": 1.7246,
+      "step": 323
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9856906806228983e-06,
+      "loss": 1.731,
+      "step": 324
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9854595177171967e-06,
+      "loss": 1.7228,
+      "step": 325
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9852265162303838e-06,
+      "loss": 1.7493,
+      "step": 326
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9849916765971713e-06,
+      "loss": 1.726,
+      "step": 327
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9847549992557035e-06,
+      "loss": 1.7292,
+      "step": 328
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9845164846475506e-06,
+      "loss": 1.7054,
+      "step": 329
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9842761332177115e-06,
+      "loss": 1.7375,
+      "step": 330
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9840339454146122e-06,
+      "loss": 1.7322,
+      "step": 331
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.983789921690105e-06,
+      "loss": 1.7115,
+      "step": 332
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.983544062499467e-06,
+      "loss": 1.7198,
+      "step": 333
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9832963683014007e-06,
+      "loss": 1.7491,
+      "step": 334
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98304683955803e-06,
+      "loss": 1.7056,
+      "step": 335
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.982795476734905e-06,
+      "loss": 1.7251,
+      "step": 336
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9825422803009937e-06,
+      "loss": 1.7582,
+      "step": 337
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9822872507286887e-06,
+      "loss": 1.7403,
+      "step": 338
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9820303884938002e-06,
+      "loss": 1.757,
+      "step": 339
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9817716940755586e-06,
+      "loss": 1.7103,
+      "step": 340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9815111679566127e-06,
+      "loss": 1.7091,
+      "step": 341
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9812488106230283e-06,
+      "loss": 1.7145,
+      "step": 342
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9809846225642884e-06,
+      "loss": 1.7072,
+      "step": 343
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9807186042732906e-06,
+      "loss": 1.7634,
+      "step": 344
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.980450756246348e-06,
+      "loss": 1.7488,
+      "step": 345
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.980181078983187e-06,
+      "loss": 1.7307,
+      "step": 346
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.979909572986948e-06,
+      "loss": 1.7243,
+      "step": 347
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9796362387641804e-06,
+      "loss": 1.7381,
+      "step": 348
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.979361076824848e-06,
+      "loss": 1.7573,
+      "step": 349
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.979084087682323e-06,
+      "loss": 1.7341,
+      "step": 350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9788052718533854e-06,
+      "loss": 1.7396,
+      "step": 351
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9785246298582257e-06,
+      "loss": 1.7074,
+      "step": 352
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.97824216222044e-06,
+      "loss": 1.7115,
+      "step": 353
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9779578694670306e-06,
+      "loss": 1.7513,
+      "step": 354
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9776717521284057e-06,
+      "loss": 1.7075,
+      "step": 355
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9773838107383767e-06,
+      "loss": 1.7747,
+      "step": 356
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.977094045834158e-06,
+      "loss": 1.7506,
+      "step": 357
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9768024579563677e-06,
+      "loss": 1.7221,
+      "step": 358
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9765090476490237e-06,
+      "loss": 1.7311,
+      "step": 359
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9762138154595447e-06,
+      "loss": 1.7278,
+      "step": 360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9759167619387473e-06,
+      "loss": 1.7219,
+      "step": 361
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9756178876408477e-06,
+      "loss": 1.7105,
+      "step": 362
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.975317193123459e-06,
+      "loss": 1.733,
+      "step": 363
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9750146789475883e-06,
+      "loss": 1.7371,
+      "step": 364
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9747103456776404e-06,
+      "loss": 1.7417,
+      "step": 365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9744041938814123e-06,
+      "loss": 1.6995,
+      "step": 366
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9740962241300947e-06,
+      "loss": 1.7091,
+      "step": 367
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.973786436998269e-06,
+      "loss": 1.7196,
+      "step": 368
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9734748330639085e-06,
+      "loss": 1.7152,
+      "step": 369
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9731614129083753e-06,
+      "loss": 1.7188,
+      "step": 370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9728461771164205e-06,
+      "loss": 1.72,
+      "step": 371
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9725291262761825e-06,
+      "loss": 1.7243,
+      "step": 372
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.972210260979186e-06,
+      "loss": 1.7543,
+      "step": 373
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9718895818203407e-06,
+      "loss": 1.7306,
+      "step": 374
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9715670893979414e-06,
+      "loss": 1.7302,
+      "step": 375
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.971242784313665e-06,
+      "loss": 1.7084,
+      "step": 376
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.97091666717257e-06,
+      "loss": 1.7319,
+      "step": 377
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9705887385830965e-06,
+      "loss": 1.6952,
+      "step": 378
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9702589991570645e-06,
+      "loss": 1.7267,
+      "step": 379
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.969927449509671e-06,
+      "loss": 1.7434,
+      "step": 380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9695940902594925e-06,
+      "loss": 1.6875,
+      "step": 381
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9692589220284796e-06,
+      "loss": 1.76,
+      "step": 382
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9689219454419587e-06,
+      "loss": 1.7135,
+      "step": 383
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.968583161128631e-06,
+      "loss": 1.7266,
+      "step": 384
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.968242569720569e-06,
+      "loss": 1.7552,
+      "step": 385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9679001718532177e-06,
+      "loss": 1.7288,
+      "step": 386
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9675559681653916e-06,
+      "loss": 1.7514,
+      "step": 387
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9672099592992747e-06,
+      "loss": 1.7223,
+      "step": 388
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9668621459004197e-06,
+      "loss": 1.7385,
+      "step": 389
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.966512528617745e-06,
+      "loss": 1.724,
+      "step": 390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.966161108103534e-06,
+      "loss": 1.7095,
+      "step": 391
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9658078850134365e-06,
+      "loss": 1.7238,
+      "step": 392
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.965452860006464e-06,
+      "loss": 1.7201,
+      "step": 393
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9650960337449887e-06,
+      "loss": 1.7237,
+      "step": 394
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9647374068947467e-06,
+      "loss": 1.7398,
+      "step": 395
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.96437698012483e-06,
+      "loss": 1.7283,
+      "step": 396
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9640147541076906e-06,
+      "loss": 1.711,
+      "step": 397
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9636507295191374e-06,
+      "loss": 1.7246,
+      "step": 398
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.963284907038334e-06,
+      "loss": 1.7099,
+      "step": 399
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9629172873477994e-06,
+      "loss": 1.7203,
+      "step": 400
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 1.7136287689208984,
+      "eval_runtime": 49.67,
+      "eval_samples_per_second": 40.266,
+      "eval_steps_per_second": 0.101,
+      "step": 400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9625478711334043e-06,
+      "loss": 1.7168,
+      "step": 401
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9621766590843727e-06,
+      "loss": 1.7308,
+      "step": 402
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.961803651893278e-06,
+      "loss": 1.7373,
+      "step": 403
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.961428850256044e-06,
+      "loss": 1.7396,
+      "step": 404
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9610522548719408e-06,
+      "loss": 1.7102,
+      "step": 405
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.960673866443586e-06,
+      "loss": 1.7376,
+      "step": 406
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.960293685676943e-06,
+      "loss": 1.7235,
+      "step": 407
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9599117132813184e-06,
+      "loss": 1.7191,
+      "step": 408
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.959527949969361e-06,
+      "loss": 1.7217,
+      "step": 409
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.959142396457063e-06,
+      "loss": 1.7116,
+      "step": 410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9587550534637545e-06,
+      "loss": 1.741,
+      "step": 411
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9583659217121045e-06,
+      "loss": 1.7122,
+      "step": 412
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9579750019281207e-06,
+      "loss": 1.7025,
+      "step": 413
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9575822948411453e-06,
+      "loss": 1.7266,
+      "step": 414
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9571878011838555e-06,
+      "loss": 1.7402,
+      "step": 415
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9567915216922623e-06,
+      "loss": 1.7149,
+      "step": 416
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.956393457105707e-06,
+      "loss": 1.7377,
+      "step": 417
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9559936081668643e-06,
+      "loss": 1.7338,
+      "step": 418
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9555919756217346e-06,
+      "loss": 1.7204,
+      "step": 419
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.955188560219648e-06,
+      "loss": 1.7338,
+      "step": 420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9547833627132605e-06,
+      "loss": 1.7119,
+      "step": 421
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.954376383858553e-06,
+      "loss": 1.7129,
+      "step": 422
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9539676244148294e-06,
+      "loss": 1.7237,
+      "step": 423
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9535570851447163e-06,
+      "loss": 1.717,
+      "step": 424
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.953144766814161e-06,
+      "loss": 1.7169,
+      "step": 425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.952730670192429e-06,
+      "loss": 1.715,
+      "step": 426
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9523147960521046e-06,
+      "loss": 1.7164,
+      "step": 427
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.951897145169088e-06,
+      "loss": 1.7035,
+      "step": 428
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.951477718322595e-06,
+      "loss": 1.7215,
+      "step": 429
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9510565162951534e-06,
+      "loss": 1.7114,
+      "step": 430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9506335398726043e-06,
+      "loss": 1.6955,
+      "step": 431
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.950208789844099e-06,
+      "loss": 1.7098,
+      "step": 432
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9497822670020965e-06,
+      "loss": 1.7029,
+      "step": 433
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9493539721423657e-06,
+      "loss": 1.7323,
+      "step": 434
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9489239060639796e-06,
+      "loss": 1.7347,
+      "step": 435
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.948492069569317e-06,
+      "loss": 1.7176,
+      "step": 436
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9480584634640595e-06,
+      "loss": 1.7083,
+      "step": 437
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9476230885571894e-06,
+      "loss": 1.7209,
+      "step": 438
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9471859456609905e-06,
+      "loss": 1.717,
+      "step": 439
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9467470355910435e-06,
+      "loss": 1.7056,
+      "step": 440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9463063591662283e-06,
+      "loss": 1.7518,
+      "step": 441
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.945863917208718e-06,
+      "loss": 1.7316,
+      "step": 442
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.945419710543981e-06,
+      "loss": 1.7242,
+      "step": 443
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.944973740000778e-06,
+      "loss": 1.725,
+      "step": 444
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9445260064111607e-06,
+      "loss": 1.7046,
+      "step": 445
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9440765106104692e-06,
+      "loss": 1.7248,
+      "step": 446
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9436252534373324e-06,
+      "loss": 1.7441,
+      "step": 447
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9431722357336657e-06,
+      "loss": 1.7252,
+      "step": 448
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.942717458344667e-06,
+      "loss": 1.7418,
+      "step": 449
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9422609221188204e-06,
+      "loss": 1.7011,
+      "step": 450
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9418026279078885e-06,
+      "loss": 1.7023,
+      "step": 451
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9413425765669164e-06,
+      "loss": 1.6934,
+      "step": 452
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9408807689542254e-06,
+      "loss": 1.7173,
+      "step": 453
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9404172059314145e-06,
+      "loss": 1.7049,
+      "step": 454
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9399518883633573e-06,
+      "loss": 1.7299,
+      "step": 455
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.939484817118202e-06,
+      "loss": 1.7171,
+      "step": 456
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9390159930673665e-06,
+      "loss": 1.6903,
+      "step": 457
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9385454170855415e-06,
+      "loss": 1.7166,
+      "step": 458
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9380730900506845e-06,
+      "loss": 1.6998,
+      "step": 459
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.93759901284402e-06,
+      "loss": 1.733,
+      "step": 460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.93712318635004e-06,
+      "loss": 1.7191,
+      "step": 461
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9366456114564963e-06,
+      "loss": 1.7104,
+      "step": 462
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9361662890544064e-06,
+      "loss": 1.7503,
+      "step": 463
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9356852200380463e-06,
+      "loss": 1.7334,
+      "step": 464
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9352024053049508e-06,
+      "loss": 1.7271,
+      "step": 465
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.934717845755912e-06,
+      "loss": 1.7075,
+      "step": 466
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.934231542294977e-06,
+      "loss": 1.7328,
+      "step": 467
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.933743495829447e-06,
+      "loss": 1.7177,
+      "step": 468
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9332537072698745e-06,
+      "loss": 1.7068,
+      "step": 469
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9327621775300633e-06,
+      "loss": 1.6906,
+      "step": 470
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.932268907527065e-06,
+      "loss": 1.7157,
+      "step": 471
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9317738981811776e-06,
+      "loss": 1.7172,
+      "step": 472
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.931277150415945e-06,
+      "loss": 1.7075,
+      "step": 473
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.930778665158154e-06,
+      "loss": 1.7149,
+      "step": 474
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.930278443337833e-06,
+      "loss": 1.724,
+      "step": 475
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.929776485888251e-06,
+      "loss": 1.7085,
+      "step": 476
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.929272793745915e-06,
+      "loss": 1.7154,
+      "step": 477
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.928767367850568e-06,
+      "loss": 1.7031,
+      "step": 478
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9282602091451876e-06,
+      "loss": 1.7284,
+      "step": 479
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9277513185759843e-06,
+      "loss": 1.7169,
+      "step": 480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.927240697092401e-06,
+      "loss": 1.7036,
+      "step": 481
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.926728345647108e-06,
+      "loss": 1.7206,
+      "step": 482
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9262142651960045e-06,
+      "loss": 1.7324,
+      "step": 483
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9256984566982156e-06,
+      "loss": 1.7171,
+      "step": 484
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.92518092111609e-06,
+      "loss": 1.7112,
+      "step": 485
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9246616594151985e-06,
+      "loss": 1.7118,
+      "step": 486
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9241406725643324e-06,
+      "loss": 1.6925,
+      "step": 487
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.923617961535502e-06,
+      "loss": 1.7114,
+      "step": 488
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9230935273039347e-06,
+      "loss": 1.7372,
+      "step": 489
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9225673708480714e-06,
+      "loss": 1.7043,
+      "step": 490
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9220394931495682e-06,
+      "loss": 1.7367,
+      "step": 491
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9215098951932905e-06,
+      "loss": 1.7057,
+      "step": 492
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.920978577967315e-06,
+      "loss": 1.7075,
+      "step": 493
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9204455424629248e-06,
+      "loss": 1.6913,
+      "step": 494
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.919910789674609e-06,
+      "loss": 1.718,
+      "step": 495
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9193743206000612e-06,
+      "loss": 1.7201,
+      "step": 496
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9188361362401774e-06,
+      "loss": 1.696,
+      "step": 497
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9182962375990527e-06,
+      "loss": 1.6942,
+      "step": 498
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.917754625683981e-06,
+      "loss": 1.7336,
+      "step": 499
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9172113015054528e-06,
+      "loss": 1.7173,
+      "step": 500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9166662660771533e-06,
+      "loss": 1.6979,
+      "step": 501
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9161195204159603e-06,
+      "loss": 1.7287,
+      "step": 502
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9155710655419417e-06,
+      "loss": 1.7282,
+      "step": 503
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.915020902478356e-06,
+      "loss": 1.6889,
+      "step": 504
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9144690322516467e-06,
+      "loss": 1.7082,
+      "step": 505
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.913915455891444e-06,
+      "loss": 1.7064,
+      "step": 506
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9133601744305595e-06,
+      "loss": 1.6933,
+      "step": 507
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.912803188904988e-06,
+      "loss": 1.7107,
+      "step": 508
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9122445003539024e-06,
+      "loss": 1.7209,
+      "step": 509
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9116841098196536e-06,
+      "loss": 1.7244,
+      "step": 510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.911122018347766e-06,
+      "loss": 1.7114,
+      "step": 511
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.910558226986941e-06,
+      "loss": 1.7232,
+      "step": 512
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.909992736789048e-06,
+      "loss": 1.7197,
+      "step": 513
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.909425548809128e-06,
+      "loss": 1.7137,
+      "step": 514
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9088566641053887e-06,
+      "loss": 1.693,
+      "step": 515
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9082860837392036e-06,
+      "loss": 1.707,
+      "step": 516
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.90771380877511e-06,
+      "loss": 1.7118,
+      "step": 517
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9071398402808071e-06,
+      "loss": 1.7253,
+      "step": 518
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.906564179327153e-06,
+      "loss": 1.7046,
+      "step": 519
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9059868269881636e-06,
+      "loss": 1.678,
+      "step": 520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9054077843410107e-06,
+      "loss": 1.7131,
+      "step": 521
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9048270524660196e-06,
+      "loss": 1.6969,
+      "step": 522
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9042446324466672e-06,
+      "loss": 1.7292,
+      "step": 523
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.90366052536958e-06,
+      "loss": 1.7166,
+      "step": 524
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9030747323245327e-06,
+      "loss": 1.687,
+      "step": 525
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.902487254404444e-06,
+      "loss": 1.7017,
+      "step": 526
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9018980927053775e-06,
+      "loss": 1.7135,
+      "step": 527
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9013072483265375e-06,
+      "loss": 1.7002,
+      "step": 528
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9007147223702685e-06,
+      "loss": 1.719,
+      "step": 529
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.900120515942051e-06,
+      "loss": 1.7177,
+      "step": 530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8995246301505018e-06,
+      "loss": 1.7048,
+      "step": 531
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8989270661073706e-06,
+      "loss": 1.6987,
+      "step": 532
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8983278249275386e-06,
+      "loss": 1.7043,
+      "step": 533
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8977269077290154e-06,
+      "loss": 1.6792,
+      "step": 534
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8971243156329378e-06,
+      "loss": 1.7053,
+      "step": 535
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8965200497635678e-06,
+      "loss": 1.71,
+      "step": 536
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8959141112482899e-06,
+      "loss": 1.7238,
+      "step": 537
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8953065012176092e-06,
+      "loss": 1.7381,
+      "step": 538
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8946972208051494e-06,
+      "loss": 1.7182,
+      "step": 539
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.894086271147651e-06,
+      "loss": 1.7142,
+      "step": 540
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8934736533849684e-06,
+      "loss": 1.7023,
+      "step": 541
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8928593686600683e-06,
+      "loss": 1.6867,
+      "step": 542
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.892243418119027e-06,
+      "loss": 1.7163,
+      "step": 543
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8916258029110303e-06,
+      "loss": 1.7069,
+      "step": 544
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8910065241883678e-06,
+      "loss": 1.694,
+      "step": 545
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.890385583106434e-06,
+      "loss": 1.7231,
+      "step": 546
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8897629808237246e-06,
+      "loss": 1.7383,
+      "step": 547
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8891387185018343e-06,
+      "loss": 1.7032,
+      "step": 548
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8885127973054555e-06,
+      "loss": 1.7326,
+      "step": 549
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.887885218402375e-06,
+      "loss": 1.7104,
+      "step": 550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8872559829634732e-06,
+      "loss": 1.7021,
+      "step": 551
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8866250921627197e-06,
+      "loss": 1.7259,
+      "step": 552
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.885992547177174e-06,
+      "loss": 1.6761,
+      "step": 553
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8853583491869817e-06,
+      "loss": 1.6947,
+      "step": 554
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8847224993753708e-06,
+      "loss": 1.7014,
+      "step": 555
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8840849989286531e-06,
+      "loss": 1.7211,
+      "step": 556
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.883445849036219e-06,
+      "loss": 1.7043,
+      "step": 557
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8828050508905365e-06,
+      "loss": 1.7075,
+      "step": 558
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8821626056871484e-06,
+      "loss": 1.7019,
+      "step": 559
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8815185146246715e-06,
+      "loss": 1.6843,
+      "step": 560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8808727789047923e-06,
+      "loss": 1.6644,
+      "step": 561
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8802253997322654e-06,
+      "loss": 1.7232,
+      "step": 562
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.879576378314913e-06,
+      "loss": 1.7138,
+      "step": 563
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.87892571586362e-06,
+      "loss": 1.7337,
+      "step": 564
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8782734135923336e-06,
+      "loss": 1.7085,
+      "step": 565
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8776194727180607e-06,
+      "loss": 1.7102,
+      "step": 566
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8769638944608644e-06,
+      "loss": 1.7179,
+      "step": 567
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8763066800438634e-06,
+      "loss": 1.7126,
+      "step": 568
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.875647830693229e-06,
+      "loss": 1.725,
+      "step": 569
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8749873476381826e-06,
+      "loss": 1.7128,
+      "step": 570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8743252321109933e-06,
+      "loss": 1.7136,
+      "step": 571
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8736614853469766e-06,
+      "loss": 1.7263,
+      "step": 572
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8729961085844912e-06,
+      "loss": 1.7086,
+      "step": 573
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8723291030649366e-06,
+      "loss": 1.6873,
+      "step": 574
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8716604700327511e-06,
+      "loss": 1.7155,
+      "step": 575
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8709902107354101e-06,
+      "loss": 1.7364,
+      "step": 576
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8703183264234226e-06,
+      "loss": 1.7216,
+      "step": 577
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.869644818350329e-06,
+      "loss": 1.7081,
+      "step": 578
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8689696877727004e-06,
+      "loss": 1.7086,
+      "step": 579
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8682929359501337e-06,
+      "loss": 1.6819,
+      "step": 580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8676145641452513e-06,
+      "loss": 1.6939,
+      "step": 581
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8669345736236981e-06,
+      "loss": 1.7018,
+      "step": 582
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8662529656541385e-06,
+      "loss": 1.7386,
+      "step": 583
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8655697415082554e-06,
+      "loss": 1.708,
+      "step": 584
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.864884902460746e-06,
+      "loss": 1.705,
+      "step": 585
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.864198449789321e-06,
+      "loss": 1.6949,
+      "step": 586
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.863510384774702e-06,
+      "loss": 1.733,
+      "step": 587
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8628207087006184e-06,
+      "loss": 1.7167,
+      "step": 588
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8621294228538047e-06,
+      "loss": 1.6918,
+      "step": 589
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8614365285239998e-06,
+      "loss": 1.6899,
+      "step": 590
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8607420270039435e-06,
+      "loss": 1.7009,
+      "step": 591
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8600459195893736e-06,
+      "loss": 1.7179,
+      "step": 592
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.859348207579024e-06,
+      "loss": 1.6813,
+      "step": 593
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.858648892274623e-06,
+      "loss": 1.6935,
+      "step": 594
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8579479749808896e-06,
+      "loss": 1.6857,
+      "step": 595
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8572454570055318e-06,
+      "loss": 1.709,
+      "step": 596
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8565413396592437e-06,
+      "loss": 1.6924,
+      "step": 597
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8558356242557041e-06,
+      "loss": 1.7113,
+      "step": 598
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8551283121115727e-06,
+      "loss": 1.7307,
+      "step": 599
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8544194045464886e-06,
+      "loss": 1.6994,
+      "step": 600
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 1.6977063417434692,
+      "eval_runtime": 52.1975,
+      "eval_samples_per_second": 38.316,
+      "eval_steps_per_second": 0.096,
+      "step": 600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.853708902883067e-06,
+      "loss": 1.701,
+      "step": 601
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.852996808446898e-06,
+      "loss": 1.7054,
+      "step": 602
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.852283122566542e-06,
+      "loss": 1.7113,
+      "step": 603
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8515678465735307e-06,
+      "loss": 1.6969,
+      "step": 604
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8508509818023605e-06,
+      "loss": 1.7057,
+      "step": 605
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.850132529590493e-06,
+      "loss": 1.714,
+      "step": 606
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8494124912783513e-06,
+      "loss": 1.7276,
+      "step": 607
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8486908682093172e-06,
+      "loss": 1.7213,
+      "step": 608
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8479676617297301e-06,
+      "loss": 1.7079,
+      "step": 609
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8472428731888833e-06,
+      "loss": 1.6997,
+      "step": 610
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8465165039390214e-06,
+      "loss": 1.7038,
+      "step": 611
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8457885553353383e-06,
+      "loss": 1.6825,
+      "step": 612
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8450590287359746e-06,
+      "loss": 1.68,
+      "step": 613
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.844327925502015e-06,
+      "loss": 1.7155,
+      "step": 614
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8435952469974856e-06,
+      "loss": 1.6959,
+      "step": 615
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8428609945893517e-06,
+      "loss": 1.7162,
+      "step": 616
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8421251696475148e-06,
+      "loss": 1.7051,
+      "step": 617
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8413877735448106e-06,
+      "loss": 1.6835,
+      "step": 618
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8406488076570059e-06,
+      "loss": 1.7295,
+      "step": 619
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8399082733627965e-06,
+      "loss": 1.7013,
+      "step": 620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8391661720438037e-06,
+      "loss": 1.7159,
+      "step": 621
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8384225050845733e-06,
+      "loss": 1.6951,
+      "step": 622
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8376772738725719e-06,
+      "loss": 1.7206,
+      "step": 623
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8369304797981842e-06,
+      "loss": 1.6984,
+      "step": 624
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.836182124254711e-06,
+      "loss": 1.7035,
+      "step": 625
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.835432208638366e-06,
+      "loss": 1.6859,
+      "step": 626
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.834680734348274e-06,
+      "loss": 1.6745,
+      "step": 627
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8339277027864682e-06,
+      "loss": 1.6968,
+      "step": 628
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8331731153578857e-06,
+      "loss": 1.7197,
+      "step": 629
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.832416973470368e-06,
+      "loss": 1.7199,
+      "step": 630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8316592785346564e-06,
+      "loss": 1.6945,
+      "step": 631
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.830900031964389e-06,
+      "loss": 1.7035,
+      "step": 632
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8301392351760992e-06,
+      "loss": 1.6744,
+      "step": 633
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8293768895892132e-06,
+      "loss": 1.6879,
+      "step": 634
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.828612996626046e-06,
+      "loss": 1.6839,
+      "step": 635
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8278475577118e-06,
+      "loss": 1.7149,
+      "step": 636
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8270805742745616e-06,
+      "loss": 1.6684,
+      "step": 637
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8263120477452998e-06,
+      "loss": 1.6789,
+      "step": 638
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8255419795578608e-06,
+      "loss": 1.7136,
+      "step": 639
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8247703711489684e-06,
+      "loss": 1.6858,
+      "step": 640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8239972239582202e-06,
+      "loss": 1.7076,
+      "step": 641
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8232225394280833e-06,
+      "loss": 1.7415,
+      "step": 642
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8224463190038948e-06,
+      "loss": 1.7141,
+      "step": 643
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8216685641338557e-06,
+      "loss": 1.7229,
+      "step": 644
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8208892762690315e-06,
+      "loss": 1.7081,
+      "step": 645
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.820108456863346e-06,
+      "loss": 1.6995,
+      "step": 646
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.819326107373582e-06,
+      "loss": 1.7146,
+      "step": 647
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8185422292593757e-06,
+      "loss": 1.6858,
+      "step": 648
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8177568239832163e-06,
+      "loss": 1.7135,
+      "step": 649
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8169698930104419e-06,
+      "loss": 1.7312,
+      "step": 650
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8161814378092368e-06,
+      "loss": 1.7179,
+      "step": 651
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8153914598506295e-06,
+      "loss": 1.7019,
+      "step": 652
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.814599960608489e-06,
+      "loss": 1.6837,
+      "step": 653
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8138069415595233e-06,
+      "loss": 1.7217,
+      "step": 654
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.813012404183275e-06,
+      "loss": 1.7006,
+      "step": 655
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8122163499621205e-06,
+      "loss": 1.6838,
+      "step": 656
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8114187803812655e-06,
+      "loss": 1.7026,
+      "step": 657
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.810619696928743e-06,
+      "loss": 1.6757,
+      "step": 658
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8098191010954106e-06,
+      "loss": 1.685,
+      "step": 659
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8090169943749474e-06,
+      "loss": 1.7231,
+      "step": 660
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8082133782638515e-06,
+      "loss": 1.7104,
+      "step": 661
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8074082542614373e-06,
+      "loss": 1.6926,
+      "step": 662
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8066016238698317e-06,
+      "loss": 1.694,
+      "step": 663
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8057934885939732e-06,
+      "loss": 1.7104,
+      "step": 664
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8049838499416068e-06,
+      "loss": 1.6976,
+      "step": 665
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.804172709423284e-06,
+      "loss": 1.6978,
+      "step": 666
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8033600685523559e-06,
+      "loss": 1.7017,
+      "step": 667
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8025459288449753e-06,
+      "loss": 1.6985,
+      "step": 668
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8017302918200898e-06,
+      "loss": 1.7021,
+      "step": 669
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8009131589994416e-06,
+      "loss": 1.6925,
+      "step": 670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8000945319075627e-06,
+      "loss": 1.6944,
+      "step": 671
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7992744120717733e-06,
+      "loss": 1.7081,
+      "step": 672
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.798452801022179e-06,
+      "loss": 1.71,
+      "step": 673
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7976297002916672e-06,
+      "loss": 1.7014,
+      "step": 674
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7968051114159043e-06,
+      "loss": 1.7191,
+      "step": 675
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7959790359333345e-06,
+      "loss": 1.717,
+      "step": 676
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7951514753851734e-06,
+      "loss": 1.7209,
+      "step": 677
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7943224313154095e-06,
+      "loss": 1.6991,
+      "step": 678
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7934919052707977e-06,
+      "loss": 1.6862,
+      "step": 679
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.792659898800858e-06,
+      "loss": 1.6934,
+      "step": 680
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7918264134578735e-06,
+      "loss": 1.6824,
+      "step": 681
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.790991450796885e-06,
+      "loss": 1.6701,
+      "step": 682
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7901550123756903e-06,
+      "loss": 1.6949,
+      "step": 683
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7893170997548407e-06,
+      "loss": 1.7421,
+      "step": 684
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7884777144976373e-06,
+      "loss": 1.6745,
+      "step": 685
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7876368581701294e-06,
+      "loss": 1.6889,
+      "step": 686
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7867945323411106e-06,
+      "loss": 1.6898,
+      "step": 687
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.785950738582116e-06,
+      "loss": 1.6949,
+      "step": 688
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7851054784674197e-06,
+      "loss": 1.6631,
+      "step": 689
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7842587535740313e-06,
+      "loss": 1.7096,
+      "step": 690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7834105654816933e-06,
+      "loss": 1.6993,
+      "step": 691
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7825609157728785e-06,
+      "loss": 1.7024,
+      "step": 692
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7817098060327863e-06,
+      "loss": 1.7222,
+      "step": 693
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7808572378493402e-06,
+      "loss": 1.7341,
+      "step": 694
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7800032128131846e-06,
+      "loss": 1.6727,
+      "step": 695
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.779147732517682e-06,
+      "loss": 1.6844,
+      "step": 696
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7782907985589103e-06,
+      "loss": 1.6949,
+      "step": 697
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.777432412535659e-06,
+      "loss": 1.7136,
+      "step": 698
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.776572576049427e-06,
+      "loss": 1.71,
+      "step": 699
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7757112907044198e-06,
+      "loss": 1.7071,
+      "step": 700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.774848558107545e-06,
+      "loss": 1.6908,
+      "step": 701
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.773984379868411e-06,
+      "loss": 1.7105,
+      "step": 702
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7731187575993239e-06,
+      "loss": 1.7222,
+      "step": 703
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7722516929152825e-06,
+      "loss": 1.7035,
+      "step": 704
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7713831874339777e-06,
+      "loss": 1.7056,
+      "step": 705
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7705132427757892e-06,
+      "loss": 1.7154,
+      "step": 706
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7696418605637802e-06,
+      "loss": 1.7195,
+      "step": 707
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7687690424236965e-06,
+      "loss": 1.7124,
+      "step": 708
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7678947899839638e-06,
+      "loss": 1.6955,
+      "step": 709
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7670191048756826e-06,
+      "loss": 1.699,
+      "step": 710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7661419887326274e-06,
+      "loss": 1.6926,
+      "step": 711
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7652634431912415e-06,
+      "loss": 1.721,
+      "step": 712
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7643834698906357e-06,
+      "loss": 1.7365,
+      "step": 713
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.763502070472585e-06,
+      "loss": 1.7193,
+      "step": 714
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.762619246581524e-06,
+      "loss": 1.7197,
+      "step": 715
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7617349998645454e-06,
+      "loss": 1.6917,
+      "step": 716
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7608493319713976e-06,
+      "loss": 1.6918,
+      "step": 717
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7599622445544787e-06,
+      "loss": 1.6763,
+      "step": 718
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7590737392688361e-06,
+      "loss": 1.6921,
+      "step": 719
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7581838177721627e-06,
+      "loss": 1.7103,
+      "step": 720
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7572924817247936e-06,
+      "loss": 1.6884,
+      "step": 721
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7563997327897025e-06,
+      "loss": 1.7176,
+      "step": 722
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7555055726324999e-06,
+      "loss": 1.696,
+      "step": 723
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7546100029214285e-06,
+      "loss": 1.7036,
+      "step": 724
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.753713025327361e-06,
+      "loss": 1.6957,
+      "step": 725
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7528146415237974e-06,
+      "loss": 1.7018,
+      "step": 726
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7519148531868609e-06,
+      "loss": 1.7104,
+      "step": 727
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7510136619952944e-06,
+      "loss": 1.6782,
+      "step": 728
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7501110696304595e-06,
+      "loss": 1.6986,
+      "step": 729
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7492070777763308e-06,
+      "loss": 1.6749,
+      "step": 730
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7483016881194948e-06,
+      "loss": 1.7237,
+      "step": 731
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7473949023491454e-06,
+      "loss": 1.699,
+      "step": 732
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7464867221570812e-06,
+      "loss": 1.7005,
+      "step": 733
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7455771492377027e-06,
+      "loss": 1.6773,
+      "step": 734
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7446661852880086e-06,
+      "loss": 1.6799,
+      "step": 735
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7437538320075928e-06,
+      "loss": 1.6931,
+      "step": 736
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.742840091098642e-06,
+      "loss": 1.69,
+      "step": 737
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7419249642659306e-06,
+      "loss": 1.6865,
+      "step": 738
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7410084532168193e-06,
+      "loss": 1.7286,
+      "step": 739
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7400905596612518e-06,
+      "loss": 1.7045,
+      "step": 740
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7391712853117508e-06,
+      "loss": 1.7035,
+      "step": 741
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7382506318834146e-06,
+      "loss": 1.6868,
+      "step": 742
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7373286010939156e-06,
+      "loss": 1.6729,
+      "step": 743
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.736405194663495e-06,
+      "loss": 1.6979,
+      "step": 744
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7354804143149608e-06,
+      "loss": 1.7198,
+      "step": 745
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7345542617736846e-06,
+      "loss": 1.6819,
+      "step": 746
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.733626738767598e-06,
+      "loss": 1.687,
+      "step": 747
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7326978470271892e-06,
+      "loss": 1.6713,
+      "step": 748
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7317675882855003e-06,
+      "loss": 1.6596,
+      "step": 749
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7308359642781241e-06,
+      "loss": 1.7147,
+      "step": 750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7299029767431998e-06,
+      "loss": 1.6864,
+      "step": 751
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7289686274214115e-06,
+      "loss": 1.7119,
+      "step": 752
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7280329180559834e-06,
+      "loss": 1.6837,
+      "step": 753
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.727095850392677e-06,
+      "loss": 1.6882,
+      "step": 754
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7261574261797885e-06,
+      "loss": 1.6978,
+      "step": 755
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.725217647168145e-06,
+      "loss": 1.6947,
+      "step": 756
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7242765151111004e-06,
+      "loss": 1.7025,
+      "step": 757
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7233340317645347e-06,
+      "loss": 1.7117,
+      "step": 758
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.722390198886847e-06,
+      "loss": 1.6973,
+      "step": 759
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7214450182389558e-06,
+      "loss": 1.7387,
+      "step": 760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.720498491584293e-06,
+      "loss": 1.7072,
+      "step": 761
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7195506206888028e-06,
+      "loss": 1.7098,
+      "step": 762
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7186014073209366e-06,
+      "loss": 1.7132,
+      "step": 763
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.717650853251651e-06,
+      "loss": 1.6926,
+      "step": 764
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7166989602544034e-06,
+      "loss": 1.6816,
+      "step": 765
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7157457301051496e-06,
+      "loss": 1.7112,
+      "step": 766
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.71479116458234e-06,
+      "loss": 1.676,
+      "step": 767
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7138352654669169e-06,
+      "loss": 1.7244,
+      "step": 768
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7128780345423098e-06,
+      "loss": 1.6801,
+      "step": 769
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7119194735944336e-06,
+      "loss": 1.7145,
+      "step": 770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7109595844116846e-06,
+      "loss": 1.6771,
+      "step": 771
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7099983687849371e-06,
+      "loss": 1.7116,
+      "step": 772
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.70903582850754e-06,
+      "loss": 1.6949,
+      "step": 773
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7080719653753142e-06,
+      "loss": 1.6751,
+      "step": 774
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7071067811865474e-06,
+      "loss": 1.7033,
+      "step": 775
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7061402777419938e-06,
+      "loss": 1.682,
+      "step": 776
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7051724568448673e-06,
+      "loss": 1.7055,
+      "step": 777
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7042033203008406e-06,
+      "loss": 1.6949,
+      "step": 778
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.703232869918041e-06,
+      "loss": 1.7107,
+      "step": 779
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7022611075070473e-06,
+      "loss": 1.6952,
+      "step": 780
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.701288034880885e-06,
+      "loss": 1.706,
+      "step": 781
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.700313653855025e-06,
+      "loss": 1.7038,
+      "step": 782
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.69933796624738e-06,
+      "loss": 1.6809,
+      "step": 783
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.698360973878299e-06,
+      "loss": 1.7146,
+      "step": 784
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6973826785705657e-06,
+      "loss": 1.7164,
+      "step": 785
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6964030821493952e-06,
+      "loss": 1.6776,
+      "step": 786
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6954221864424294e-06,
+      "loss": 1.6978,
+      "step": 787
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6944399932797351e-06,
+      "loss": 1.7013,
+      "step": 788
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.693456504493799e-06,
+      "loss": 1.6988,
+      "step": 789
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6924717219195257e-06,
+      "loss": 1.7197,
+      "step": 790
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.691485647394233e-06,
+      "loss": 1.6934,
+      "step": 791
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6904982827576497e-06,
+      "loss": 1.6853,
+      "step": 792
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6895096298519112e-06,
+      "loss": 1.7044,
+      "step": 793
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6885196905215566e-06,
+      "loss": 1.6847,
+      "step": 794
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.687528466613525e-06,
+      "loss": 1.6964,
+      "step": 795
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.686535959977152e-06,
+      "loss": 1.6891,
+      "step": 796
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6855421724641672e-06,
+      "loss": 1.6853,
+      "step": 797
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6845471059286886e-06,
+      "loss": 1.7069,
+      "step": 798
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6835507622272216e-06,
+      "loss": 1.7098,
+      "step": 799
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.682553143218654e-06,
+      "loss": 1.6884,
+      "step": 800
+    },
+    {
+      "epoch": 0.56,
+      "eval_loss": 1.6870813369750977,
+      "eval_runtime": 51.8684,
+      "eval_samples_per_second": 38.559,
+      "eval_steps_per_second": 0.096,
+      "step": 800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.681554250764253e-06,
+      "loss": 1.7207,
+      "step": 801
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6805540867276618e-06,
+      "loss": 1.6815,
+      "step": 802
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.679552652974895e-06,
+      "loss": 1.6979,
+      "step": 803
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.678549951374338e-06,
+      "loss": 1.6991,
+      "step": 804
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6775459837967409e-06,
+      "loss": 1.7017,
+      "step": 805
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6765407521152145e-06,
+      "loss": 1.6996,
+      "step": 806
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6755342582052292e-06,
+      "loss": 1.7013,
+      "step": 807
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6745265039446106e-06,
+      "loss": 1.701,
+      "step": 808
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.673517491213535e-06,
+      "loss": 1.6765,
+      "step": 809
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.672507221894527e-06,
+      "loss": 1.6883,
+      "step": 810
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.671495697872456e-06,
+      "loss": 1.662,
+      "step": 811
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6704829210345315e-06,
+      "loss": 1.7141,
+      "step": 812
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6694688932703005e-06,
+      "loss": 1.6866,
+      "step": 813
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.668453616471645e-06,
+      "loss": 1.6929,
+      "step": 814
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6674370925327757e-06,
+      "loss": 1.6792,
+      "step": 815
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6664193233502313e-06,
+      "loss": 1.7023,
+      "step": 816
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6654003108228733e-06,
+      "loss": 1.7033,
+      "step": 817
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.664380056851883e-06,
+      "loss": 1.6852,
+      "step": 818
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6633585633407577e-06,
+      "loss": 1.7174,
+      "step": 819
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6623358321953075e-06,
+      "loss": 1.6792,
+      "step": 820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6613118653236517e-06,
+      "loss": 1.6746,
+      "step": 821
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.660286664636215e-06,
+      "loss": 1.6905,
+      "step": 822
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6592602320457236e-06,
+      "loss": 1.7006,
+      "step": 823
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.658232569467203e-06,
+      "loss": 1.6985,
+      "step": 824
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6572036788179725e-06,
+      "loss": 1.6838,
+      "step": 825
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6561735620176432e-06,
+      "loss": 1.6922,
+      "step": 826
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6551422209881136e-06,
+      "loss": 1.6903,
+      "step": 827
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.654109657653567e-06,
+      "loss": 1.6983,
+      "step": 828
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6530758739404655e-06,
+      "loss": 1.6818,
+      "step": 829
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6520408717775506e-06,
+      "loss": 1.7134,
+      "step": 830
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.651004653095834e-06,
+      "loss": 1.6885,
+      "step": 831
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6499672198285995e-06,
+      "loss": 1.7069,
+      "step": 832
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.648928573911396e-06,
+      "loss": 1.7036,
+      "step": 833
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.647888717282035e-06,
+      "loss": 1.6838,
+      "step": 834
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.646847651880587e-06,
+      "loss": 1.7183,
+      "step": 835
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6458053796493769e-06,
+      "loss": 1.7012,
+      "step": 836
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6447619025329825e-06,
+      "loss": 1.7206,
+      "step": 837
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6437172224782287e-06,
+      "loss": 1.6852,
+      "step": 838
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.642671341434185e-06,
+      "loss": 1.7036,
+      "step": 839
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6416242613521608e-06,
+      "loss": 1.6805,
+      "step": 840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6405759841857045e-06,
+      "loss": 1.6983,
+      "step": 841
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.639526511890596e-06,
+      "loss": 1.693,
+      "step": 842
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6384758464248454e-06,
+      "loss": 1.6935,
+      "step": 843
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6374239897486897e-06,
+      "loss": 1.6884,
+      "step": 844
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6363709438245877e-06,
+      "loss": 1.6931,
+      "step": 845
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6353167106172165e-06,
+      "loss": 1.6818,
+      "step": 846
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6342612920934702e-06,
+      "loss": 1.7077,
+      "step": 847
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6332046902224518e-06,
+      "loss": 1.6893,
+      "step": 848
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6321469069754738e-06,
+      "loss": 1.6643,
+      "step": 849
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6310879443260529e-06,
+      "loss": 1.6865,
+      "step": 850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.630027804249905e-06,
+      "loss": 1.6885,
+      "step": 851
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6289664887249437e-06,
+      "loss": 1.717,
+      "step": 852
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6279039997312754e-06,
+      "loss": 1.6804,
+      "step": 853
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.626840339251196e-06,
+      "loss": 1.6942,
+      "step": 854
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6257755092691862e-06,
+      "loss": 1.6848,
+      "step": 855
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6247095117719102e-06,
+      "loss": 1.6965,
+      "step": 856
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6236423487482093e-06,
+      "loss": 1.69,
+      "step": 857
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6225740221890996e-06,
+      "loss": 1.6837,
+      "step": 858
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6215045340877679e-06,
+      "loss": 1.6942,
+      "step": 859
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.620433886439568e-06,
+      "loss": 1.6624,
+      "step": 860
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6193620812420177e-06,
+      "loss": 1.7134,
+      "step": 861
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6182891204947942e-06,
+      "loss": 1.693,
+      "step": 862
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.61721500619973e-06,
+      "loss": 1.6747,
+      "step": 863
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6161397403608109e-06,
+      "loss": 1.6863,
+      "step": 864
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6150633249841694e-06,
+      "loss": 1.6767,
+      "step": 865
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.613985762078085e-06,
+      "loss": 1.6915,
+      "step": 866
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6129070536529765e-06,
+      "loss": 1.6798,
+      "step": 867
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6118272017214002e-06,
+      "loss": 1.7064,
+      "step": 868
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6107462082980465e-06,
+      "loss": 1.6899,
+      "step": 869
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6096640753997346e-06,
+      "loss": 1.667,
+      "step": 870
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.60858080504541e-06,
+      "loss": 1.6919,
+      "step": 871
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6074963992561407e-06,
+      "loss": 1.6891,
+      "step": 872
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6064108600551128e-06,
+      "loss": 1.6804,
+      "step": 873
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6053241894676269e-06,
+      "loss": 1.7016,
+      "step": 874
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6042363895210945e-06,
+      "loss": 1.7014,
+      "step": 875
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6031474622450343e-06,
+      "loss": 1.7025,
+      "step": 876
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6020574096710683e-06,
+      "loss": 1.7113,
+      "step": 877
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6009662338329176e-06,
+      "loss": 1.7095,
+      "step": 878
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5998739367663995e-06,
+      "loss": 1.6943,
+      "step": 879
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5987805205094225e-06,
+      "loss": 1.6797,
+      "step": 880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5976859871019843e-06,
+      "loss": 1.7214,
+      "step": 881
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5965903385861656e-06,
+      "loss": 1.7275,
+      "step": 882
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5954935770061284e-06,
+      "loss": 1.7033,
+      "step": 883
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5943957044081113e-06,
+      "loss": 1.6811,
+      "step": 884
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5932967228404253e-06,
+      "loss": 1.6751,
+      "step": 885
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5921966343534506e-06,
+      "loss": 1.7052,
+      "step": 886
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5910954409996328e-06,
+      "loss": 1.7037,
+      "step": 887
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5899931448334787e-06,
+      "loss": 1.6678,
+      "step": 888
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5888897479115528e-06,
+      "loss": 1.7323,
+      "step": 889
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.587785252292473e-06,
+      "loss": 1.6831,
+      "step": 890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5866796600369072e-06,
+      "loss": 1.7182,
+      "step": 891
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5855729732075692e-06,
+      "loss": 1.705,
+      "step": 892
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5844651938692156e-06,
+      "loss": 1.6758,
+      "step": 893
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5833563240886396e-06,
+      "loss": 1.7021,
+      "step": 894
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.582246365934671e-06,
+      "loss": 1.6791,
+      "step": 895
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5811353214781688e-06,
+      "loss": 1.669,
+      "step": 896
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5800231927920194e-06,
+      "loss": 1.6927,
+      "step": 897
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5789099819511313e-06,
+      "loss": 1.6966,
+      "step": 898
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.577795691032433e-06,
+      "loss": 1.6932,
+      "step": 899
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5766803221148673e-06,
+      "loss": 1.7078,
+      "step": 900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5755638772793882e-06,
+      "loss": 1.6804,
+      "step": 901
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5744463586089578e-06,
+      "loss": 1.6672,
+      "step": 902
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.573327768188541e-06,
+      "loss": 1.6922,
+      "step": 903
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5722081081051027e-06,
+      "loss": 1.6869,
+      "step": 904
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5710873804476032e-06,
+      "loss": 1.6968,
+      "step": 905
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5699655873069946e-06,
+      "loss": 1.685,
+      "step": 906
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5688427307762169e-06,
+      "loss": 1.7052,
+      "step": 907
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5677188129501938e-06,
+      "loss": 1.6938,
+      "step": 908
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5665938359258305e-06,
+      "loss": 1.6796,
+      "step": 909
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.565467801802006e-06,
+      "loss": 1.675,
+      "step": 910
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5643407126795731e-06,
+      "loss": 1.714,
+      "step": 911
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.563212570661353e-06,
+      "loss": 1.6874,
+      "step": 912
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5620833778521306e-06,
+      "loss": 1.7197,
+      "step": 913
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5609531363586515e-06,
+      "loss": 1.6764,
+      "step": 914
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.559821848289618e-06,
+      "loss": 1.679,
+      "step": 915
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5586895157556853e-06,
+      "loss": 1.6845,
+      "step": 916
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5575561408694563e-06,
+      "loss": 1.6999,
+      "step": 917
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5564217257454794e-06,
+      "loss": 1.7045,
+      "step": 918
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5552862725002436e-06,
+      "loss": 1.6781,
+      "step": 919
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5541497832521748e-06,
+      "loss": 1.6769,
+      "step": 920
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5530122601216316e-06,
+      "loss": 1.6793,
+      "step": 921
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5518737052309017e-06,
+      "loss": 1.71,
+      "step": 922
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.550734120704198e-06,
+      "loss": 1.6999,
+      "step": 923
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.549593508667653e-06,
+      "loss": 1.7036,
+      "step": 924
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5484518712493187e-06,
+      "loss": 1.6987,
+      "step": 925
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.547309210579158e-06,
+      "loss": 1.7145,
+      "step": 926
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5461655287890438e-06,
+      "loss": 1.6744,
+      "step": 927
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5450208280127542e-06,
+      "loss": 1.6855,
+      "step": 928
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5438751103859682e-06,
+      "loss": 1.6876,
+      "step": 929
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5427283780462616e-06,
+      "loss": 1.7037,
+      "step": 930
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.541580633133104e-06,
+      "loss": 1.6871,
+      "step": 931
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5404318777878542e-06,
+      "loss": 1.7001,
+      "step": 932
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5392821141537555e-06,
+      "loss": 1.695,
+      "step": 933
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5381313443759324e-06,
+      "loss": 1.7151,
+      "step": 934
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.536979570601388e-06,
+      "loss": 1.6975,
+      "step": 935
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5358267949789964e-06,
+      "loss": 1.6651,
+      "step": 936
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5346730196595028e-06,
+      "loss": 1.6857,
+      "step": 937
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5335182467955161e-06,
+      "loss": 1.6761,
+      "step": 938
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5323624785415069e-06,
+      "loss": 1.672,
+      "step": 939
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5312057170538033e-06,
+      "loss": 1.6414,
+      "step": 940
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5300479644905861e-06,
+      "loss": 1.7188,
+      "step": 941
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5288892230118846e-06,
+      "loss": 1.7076,
+      "step": 942
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5277294947795744e-06,
+      "loss": 1.6866,
+      "step": 943
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.526568781957371e-06,
+      "loss": 1.6909,
+      "step": 944
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5254070867108274e-06,
+      "loss": 1.6833,
+      "step": 945
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5242444112073293e-06,
+      "loss": 1.6878,
+      "step": 946
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5230807576160914e-06,
+      "loss": 1.6596,
+      "step": 947
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5219161281081535e-06,
+      "loss": 1.6715,
+      "step": 948
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5207505248563753e-06,
+      "loss": 1.6972,
+      "step": 949
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5195839500354335e-06,
+      "loss": 1.6771,
+      "step": 950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5184164058218185e-06,
+      "loss": 1.6972,
+      "step": 951
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5172478943938285e-06,
+      "loss": 1.6911,
+      "step": 952
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.516078417931566e-06,
+      "loss": 1.6834,
+      "step": 953
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5149079786169341e-06,
+      "loss": 1.6881,
+      "step": 954
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5137365786336327e-06,
+      "loss": 1.6971,
+      "step": 955
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.512564220167154e-06,
+      "loss": 1.689,
+      "step": 956
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5113909054047778e-06,
+      "loss": 1.68,
+      "step": 957
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5102166365355687e-06,
+      "loss": 1.6699,
+      "step": 958
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5090414157503713e-06,
+      "loss": 1.688,
+      "step": 959
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5078652452418061e-06,
+      "loss": 1.7217,
+      "step": 960
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5066881272042653e-06,
+      "loss": 1.6949,
+      "step": 961
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.505510063833909e-06,
+      "loss": 1.6872,
+      "step": 962
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5043310573286617e-06,
+      "loss": 1.6974,
+      "step": 963
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5031511098882069e-06,
+      "loss": 1.7089,
+      "step": 964
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5019702237139828e-06,
+      "loss": 1.6834,
+      "step": 965
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5007884010091806e-06,
+      "loss": 1.666,
+      "step": 966
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4996056439787381e-06,
+      "loss": 1.7039,
+      "step": 967
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.498421954829336e-06,
+      "loss": 1.6851,
+      "step": 968
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4972373357693944e-06,
+      "loss": 1.7099,
+      "step": 969
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.496051789009068e-06,
+      "loss": 1.6728,
+      "step": 970
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.494865316760243e-06,
+      "loss": 1.6966,
+      "step": 971
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4936779212365317e-06,
+      "loss": 1.7122,
+      "step": 972
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4924896046532687e-06,
+      "loss": 1.6933,
+      "step": 973
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4913003692275082e-06,
+      "loss": 1.6778,
+      "step": 974
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4901102171780174e-06,
+      "loss": 1.6918,
+      "step": 975
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4889191507252741e-06,
+      "loss": 1.696,
+      "step": 976
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4877271720914629e-06,
+      "loss": 1.6746,
+      "step": 977
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4865342835004686e-06,
+      "loss": 1.7027,
+      "step": 978
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4853404871778757e-06,
+      "loss": 1.6893,
+      "step": 979
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4841457853509606e-06,
+      "loss": 1.6804,
+      "step": 980
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4829501802486897e-06,
+      "loss": 1.7041,
+      "step": 981
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4817536741017151e-06,
+      "loss": 1.682,
+      "step": 982
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4805562691423697e-06,
+      "loss": 1.6706,
+      "step": 983
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4793579676046628e-06,
+      "loss": 1.6845,
+      "step": 984
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4781587717242768e-06,
+      "loss": 1.6842,
+      "step": 985
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4769586837385633e-06,
+      "loss": 1.7011,
+      "step": 986
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4757577058865374e-06,
+      "loss": 1.7018,
+      "step": 987
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.474555840408875e-06,
+      "loss": 1.7038,
+      "step": 988
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4733530895479072e-06,
+      "loss": 1.6913,
+      "step": 989
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4721494555476187e-06,
+      "loss": 1.675,
+      "step": 990
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4709449406536398e-06,
+      "loss": 1.6841,
+      "step": 991
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4697395471132458e-06,
+      "loss": 1.6973,
+      "step": 992
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4685332771753508e-06,
+      "loss": 1.6911,
+      "step": 993
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.467326133090504e-06,
+      "loss": 1.6784,
+      "step": 994
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4661181171108853e-06,
+      "loss": 1.6816,
+      "step": 995
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4649092314903015e-06,
+      "loss": 1.6426,
+      "step": 996
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4636994784841823e-06,
+      "loss": 1.6916,
+      "step": 997
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.462488860349575e-06,
+      "loss": 1.6596,
+      "step": 998
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.461277379345141e-06,
+      "loss": 1.7094,
+      "step": 999
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.460065037731152e-06,
+      "loss": 1.6836,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7,
+      "eval_loss": 1.67930269241333,
+      "eval_runtime": 52.0531,
+      "eval_samples_per_second": 38.422,
+      "eval_steps_per_second": 0.096,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4588518377694854e-06,
+      "loss": 1.6567,
+      "step": 1001
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.45763778172362e-06,
+      "loss": 1.6634,
+      "step": 1002
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.456422871858631e-06,
+      "loss": 1.6733,
+      "step": 1003
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.455207110441187e-06,
+      "loss": 1.6626,
+      "step": 1004
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4539904997395467e-06,
+      "loss": 1.6891,
+      "step": 1005
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4527730420235512e-06,
+      "loss": 1.7102,
+      "step": 1006
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.451554739564623e-06,
+      "loss": 1.7025,
+      "step": 1007
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.450335594635761e-06,
+      "loss": 1.6704,
+      "step": 1008
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4491156095115344e-06,
+      "loss": 1.682,
+      "step": 1009
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.447894786468082e-06,
+      "loss": 1.7065,
+      "step": 1010
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.446673127783104e-06,
+      "loss": 1.7041,
+      "step": 1011
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4454506357358609e-06,
+      "loss": 1.7085,
+      "step": 1012
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.444227312607168e-06,
+      "loss": 1.6725,
+      "step": 1013
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4430031606793901e-06,
+      "loss": 1.6527,
+      "step": 1014
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4417781822364395e-06,
+      "loss": 1.6583,
+      "step": 1015
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4405523795637696e-06,
+      "loss": 1.684,
+      "step": 1016
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4393257549483726e-06,
+      "loss": 1.6759,
+      "step": 1017
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4380983106787729e-06,
+      "loss": 1.6884,
+      "step": 1018
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4368700490450254e-06,
+      "loss": 1.6843,
+      "step": 1019
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.435640972338709e-06,
+      "loss": 1.681,
+      "step": 1020
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4344110828529244e-06,
+      "loss": 1.6913,
+      "step": 1021
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.433180382882287e-06,
+      "loss": 1.6948,
+      "step": 1022
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.431948874722926e-06,
+      "loss": 1.7169,
+      "step": 1023
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4307165606724775e-06,
+      "loss": 1.6975,
+      "step": 1024
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.429483443030082e-06,
+      "loss": 1.6851,
+      "step": 1025
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.428249524096378e-06,
+      "loss": 1.6912,
+      "step": 1026
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4270148061734996e-06,
+      "loss": 1.7128,
+      "step": 1027
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4257792915650725e-06,
+      "loss": 1.7199,
+      "step": 1028
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4245429825762076e-06,
+      "loss": 1.6807,
+      "step": 1029
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4233058815134977e-06,
+      "loss": 1.6831,
+      "step": 1030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4220679906850146e-06,
+      "loss": 1.6927,
+      "step": 1031
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4208293124003025e-06,
+      "loss": 1.6953,
+      "step": 1032
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4195898489703755e-06,
+      "loss": 1.7036,
+      "step": 1033
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4183496027077117e-06,
+      "loss": 1.706,
+      "step": 1034
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4171085759262505e-06,
+      "loss": 1.6825,
+      "step": 1035
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4158667709413874e-06,
+      "loss": 1.6546,
+      "step": 1036
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4146241900699698e-06,
+      "loss": 1.6963,
+      "step": 1037
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.413380835630292e-06,
+      "loss": 1.6876,
+      "step": 1038
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4121367099420923e-06,
+      "loss": 1.7079,
+      "step": 1039
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4108918153265483e-06,
+      "loss": 1.6693,
+      "step": 1040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4096461541062708e-06,
+      "loss": 1.7173,
+      "step": 1041
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.408399728605302e-06,
+      "loss": 1.7011,
+      "step": 1042
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4071525411491097e-06,
+      "loss": 1.6756,
+      "step": 1043
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4059045940645832e-06,
+      "loss": 1.6578,
+      "step": 1044
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4046558896800299e-06,
+      "loss": 1.6543,
+      "step": 1045
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4034064303251681e-06,
+      "loss": 1.678,
+      "step": 1046
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4021562183311267e-06,
+      "loss": 1.7089,
+      "step": 1047
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4009052560304377e-06,
+      "loss": 1.6837,
+      "step": 1048
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3996535457570332e-06,
+      "loss": 1.6942,
+      "step": 1049
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3984010898462415e-06,
+      "loss": 1.6873,
+      "step": 1050
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3971478906347805e-06,
+      "loss": 1.6982,
+      "step": 1051
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3958939504607564e-06,
+      "loss": 1.6783,
+      "step": 1052
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.394639271663657e-06,
+      "loss": 1.6774,
+      "step": 1053
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3933838565843483e-06,
+      "loss": 1.6858,
+      "step": 1054
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.39212770756507e-06,
+      "loss": 1.7111,
+      "step": 1055
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3908708269494315e-06,
+      "loss": 1.6711,
+      "step": 1056
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3896132170824063e-06,
+      "loss": 1.6611,
+      "step": 1057
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3883548803103295e-06,
+      "loss": 1.6848,
+      "step": 1058
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.387095818980892e-06,
+      "loss": 1.6943,
+      "step": 1059
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3858360354431353e-06,
+      "loss": 1.7138,
+      "step": 1060
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3845755320474504e-06,
+      "loss": 1.6651,
+      "step": 1061
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3833143111455703e-06,
+      "loss": 1.6949,
+      "step": 1062
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3820523750905667e-06,
+      "loss": 1.665,
+      "step": 1063
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.380789726236845e-06,
+      "loss": 1.6832,
+      "step": 1064
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3795263669401418e-06,
+      "loss": 1.6762,
+      "step": 1065
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3782622995575179e-06,
+      "loss": 1.6995,
+      "step": 1066
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3769975264473565e-06,
+      "loss": 1.6704,
+      "step": 1067
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3757320499693559e-06,
+      "loss": 1.6667,
+      "step": 1068
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3744658724845274e-06,
+      "loss": 1.6841,
+      "step": 1069
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3731989963551914e-06,
+      "loss": 1.6633,
+      "step": 1070
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3719314239449695e-06,
+      "loss": 1.7047,
+      "step": 1071
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.370663157618784e-06,
+      "loss": 1.6627,
+      "step": 1072
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3693941997428513e-06,
+      "loss": 1.6713,
+      "step": 1073
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3681245526846781e-06,
+      "loss": 1.697,
+      "step": 1074
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3668542188130565e-06,
+      "loss": 1.6905,
+      "step": 1075
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3655832004980608e-06,
+      "loss": 1.7147,
+      "step": 1076
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3643115001110416e-06,
+      "loss": 1.67,
+      "step": 1077
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3630391200246228e-06,
+      "loss": 1.6994,
+      "step": 1078
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3617660626126955e-06,
+      "loss": 1.6691,
+      "step": 1079
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3604923302504146e-06,
+      "loss": 1.6643,
+      "step": 1080
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3592179253141954e-06,
+      "loss": 1.6954,
+      "step": 1081
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3579428501817065e-06,
+      "loss": 1.6861,
+      "step": 1082
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3566671072318675e-06,
+      "loss": 1.6676,
+      "step": 1083
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.355390698844845e-06,
+      "loss": 1.7083,
+      "step": 1084
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3541136274020451e-06,
+      "loss": 1.6715,
+      "step": 1085
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3528358952861128e-06,
+      "loss": 1.6982,
+      "step": 1086
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3515575048809245e-06,
+      "loss": 1.688,
+      "step": 1087
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3502784585715853e-06,
+      "loss": 1.6849,
+      "step": 1088
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.348998758744424e-06,
+      "loss": 1.711,
+      "step": 1089
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3477184077869889e-06,
+      "loss": 1.6905,
+      "step": 1090
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3464374080880422e-06,
+      "loss": 1.6662,
+      "step": 1091
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3451557620375575e-06,
+      "loss": 1.7093,
+      "step": 1092
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3438734720267144e-06,
+      "loss": 1.6583,
+      "step": 1093
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3425905404478929e-06,
+      "loss": 1.6977,
+      "step": 1094
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3413069696946703e-06,
+      "loss": 1.6656,
+      "step": 1095
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.340022762161817e-06,
+      "loss": 1.6951,
+      "step": 1096
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3387379202452915e-06,
+      "loss": 1.7047,
+      "step": 1097
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3374524463422347e-06,
+      "loss": 1.6749,
+      "step": 1098
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3361663428509677e-06,
+      "loss": 1.6817,
+      "step": 1099
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.334879612170986e-06,
+      "loss": 1.696,
+      "step": 1100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3335922567029553e-06,
+      "loss": 1.6968,
+      "step": 1101
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3323042788487065e-06,
+      "loss": 1.6908,
+      "step": 1102
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3310156810112322e-06,
+      "loss": 1.7031,
+      "step": 1103
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3297264655946814e-06,
+      "loss": 1.6602,
+      "step": 1104
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3284366350043557e-06,
+      "loss": 1.6869,
+      "step": 1105
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3271461916467036e-06,
+      "loss": 1.6565,
+      "step": 1106
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3258551379293183e-06,
+      "loss": 1.6862,
+      "step": 1107
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3245634762609305e-06,
+      "loss": 1.7122,
+      "step": 1108
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3232712090514055e-06,
+      "loss": 1.7219,
+      "step": 1109
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3219783387117383e-06,
+      "loss": 1.6989,
+      "step": 1110
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3206848676540494e-06,
+      "loss": 1.6967,
+      "step": 1111
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.31939079829158e-06,
+      "loss": 1.6794,
+      "step": 1112
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3180961330386876e-06,
+      "loss": 1.6831,
+      "step": 1113
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.316800874310841e-06,
+      "loss": 1.6795,
+      "step": 1114
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3155050245246168e-06,
+      "loss": 1.7109,
+      "step": 1115
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3142085860976945e-06,
+      "loss": 1.6639,
+      "step": 1116
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.312911561448851e-06,
+      "loss": 1.6919,
+      "step": 1117
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3116139529979578e-06,
+      "loss": 1.6827,
+      "step": 1118
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3103157631659747e-06,
+      "loss": 1.7034,
+      "step": 1119
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3090169943749473e-06,
+      "loss": 1.6819,
+      "step": 1120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3077176490480007e-06,
+      "loss": 1.6847,
+      "step": 1121
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3064177296093353e-06,
+      "loss": 1.7037,
+      "step": 1122
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.305117238484223e-06,
+      "loss": 1.6962,
+      "step": 1123
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3038161780990033e-06,
+      "loss": 1.7082,
+      "step": 1124
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3025145508810757e-06,
+      "loss": 1.7038,
+      "step": 1125
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.301212359258899e-06,
+      "loss": 1.6915,
+      "step": 1126
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2999096056619842e-06,
+      "loss": 1.6768,
+      "step": 1127
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.298606292520891e-06,
+      "loss": 1.6869,
+      "step": 1128
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2973024222672232e-06,
+      "loss": 1.6824,
+      "step": 1129
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2959979973336234e-06,
+      "loss": 1.6588,
+      "step": 1130
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2946930201537699e-06,
+      "loss": 1.6649,
+      "step": 1131
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2933874931623705e-06,
+      "loss": 1.6784,
+      "step": 1132
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2920814187951596e-06,
+      "loss": 1.7014,
+      "step": 1133
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2907747994888923e-06,
+      "loss": 1.71,
+      "step": 1134
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2894676376813407e-06,
+      "loss": 1.6748,
+      "step": 1135
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2881599358112886e-06,
+      "loss": 1.6968,
+      "step": 1136
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2868516963185278e-06,
+      "loss": 1.6476,
+      "step": 1137
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2855429216438536e-06,
+      "loss": 1.6986,
+      "step": 1138
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2842336142290586e-06,
+      "loss": 1.6536,
+      "step": 1139
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.28292377651693e-06,
+      "loss": 1.6715,
+      "step": 1140
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2816134109512446e-06,
+      "loss": 1.6612,
+      "step": 1141
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2803025199767636e-06,
+      "loss": 1.6817,
+      "step": 1142
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2789911060392294e-06,
+      "loss": 1.6855,
+      "step": 1143
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2776791715853582e-06,
+      "loss": 1.6738,
+      "step": 1144
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2763667190628389e-06,
+      "loss": 1.7003,
+      "step": 1145
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.275053750920327e-06,
+      "loss": 1.6984,
+      "step": 1146
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2737402696074391e-06,
+      "loss": 1.6794,
+      "step": 1147
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2724262775747496e-06,
+      "loss": 1.688,
+      "step": 1148
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2711117772737863e-06,
+      "loss": 1.6669,
+      "step": 1149
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2697967711570242e-06,
+      "loss": 1.6787,
+      "step": 1150
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2684812616778831e-06,
+      "loss": 1.7169,
+      "step": 1151
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2671652512907212e-06,
+      "loss": 1.6912,
+      "step": 1152
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2658487424508312e-06,
+      "loss": 1.7178,
+      "step": 1153
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2645317376144366e-06,
+      "loss": 1.7027,
+      "step": 1154
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2632142392386857e-06,
+      "loss": 1.7157,
+      "step": 1155
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.261896249781647e-06,
+      "loss": 1.6758,
+      "step": 1156
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2605777717023068e-06,
+      "loss": 1.6833,
+      "step": 1157
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2592588074605612e-06,
+      "loss": 1.6819,
+      "step": 1158
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2579393595172146e-06,
+      "loss": 1.6756,
+      "step": 1159
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2566194303339736e-06,
+      "loss": 1.6807,
+      "step": 1160
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2552990223734423e-06,
+      "loss": 1.6888,
+      "step": 1161
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2539781380991186e-06,
+      "loss": 1.7061,
+      "step": 1162
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2526567799753882e-06,
+      "loss": 1.6901,
+      "step": 1163
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2513349504675219e-06,
+      "loss": 1.6895,
+      "step": 1164
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.250012652041669e-06,
+      "loss": 1.7003,
+      "step": 1165
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2486898871648551e-06,
+      "loss": 1.6856,
+      "step": 1166
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2473666583049737e-06,
+      "loss": 1.691,
+      "step": 1167
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2460429679307862e-06,
+      "loss": 1.6947,
+      "step": 1168
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.244718818511914e-06,
+      "loss": 1.6778,
+      "step": 1169
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2433942125188356e-06,
+      "loss": 1.6945,
+      "step": 1170
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2420691524228804e-06,
+      "loss": 1.6774,
+      "step": 1171
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2407436406962257e-06,
+      "loss": 1.6758,
+      "step": 1172
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2394176798118913e-06,
+      "loss": 1.7056,
+      "step": 1173
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2380912722437347e-06,
+      "loss": 1.6751,
+      "step": 1174
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2367644204664467e-06,
+      "loss": 1.6588,
+      "step": 1175
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2354371269555477e-06,
+      "loss": 1.6897,
+      "step": 1176
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2341093941873818e-06,
+      "loss": 1.6666,
+      "step": 1177
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2327812246391124e-06,
+      "loss": 1.6984,
+      "step": 1178
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2314526207887175e-06,
+      "loss": 1.6979,
+      "step": 1179
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2301235851149864e-06,
+      "loss": 1.6835,
+      "step": 1180
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2287941200975134e-06,
+      "loss": 1.6625,
+      "step": 1181
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2274642282166934e-06,
+      "loss": 1.6965,
+      "step": 1182
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.226133911953719e-06,
+      "loss": 1.6875,
+      "step": 1183
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.224803173790573e-06,
+      "loss": 1.6847,
+      "step": 1184
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2234720162100268e-06,
+      "loss": 1.6777,
+      "step": 1185
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2221404416956336e-06,
+      "loss": 1.6768,
+      "step": 1186
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2208084527317238e-06,
+      "loss": 1.6918,
+      "step": 1187
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2194760518034027e-06,
+      "loss": 1.6542,
+      "step": 1188
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2181432413965426e-06,
+      "loss": 1.6893,
+      "step": 1189
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2168100239977809e-06,
+      "loss": 1.7025,
+      "step": 1190
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2154764020945134e-06,
+      "loss": 1.6794,
+      "step": 1191
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2141423781748913e-06,
+      "loss": 1.6484,
+      "step": 1192
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2128079547278155e-06,
+      "loss": 1.6854,
+      "step": 1193
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.211473134242933e-06,
+      "loss": 1.7082,
+      "step": 1194
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.21013791921063e-06,
+      "loss": 1.6982,
+      "step": 1195
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2088023121220304e-06,
+      "loss": 1.6736,
+      "step": 1196
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.207466315468989e-06,
+      "loss": 1.6768,
+      "step": 1197
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.206129931744087e-06,
+      "loss": 1.6805,
+      "step": 1198
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2047931634406278e-06,
+      "loss": 1.6842,
+      "step": 1199
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.203456013052634e-06,
+      "loss": 1.6744,
+      "step": 1200
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 1.6732972860336304,
+      "eval_runtime": 65.2828,
+      "eval_samples_per_second": 30.636,
+      "eval_steps_per_second": 0.077,
+      "step": 1200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.202118483074838e-06,
+      "loss": 1.6948,
+      "step": 1201
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.200780576002683e-06,
+      "loss": 1.7231,
+      "step": 1202
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1994422943323148e-06,
+      "loss": 1.6662,
+      "step": 1203
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.198103640560578e-06,
+      "loss": 1.6658,
+      "step": 1204
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1967646171850117e-06,
+      "loss": 1.6802,
+      "step": 1205
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1954252267038438e-06,
+      "loss": 1.6962,
+      "step": 1206
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1940854716159887e-06,
+      "loss": 1.6758,
+      "step": 1207
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1927453544210396e-06,
+      "loss": 1.672,
+      "step": 1208
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1914048776192655e-06,
+      "loss": 1.7002,
+      "step": 1209
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1900640437116072e-06,
+      "loss": 1.6985,
+      "step": 1210
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1887228551996708e-06,
+      "loss": 1.6897,
+      "step": 1211
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1873813145857248e-06,
+      "loss": 1.7101,
+      "step": 1212
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1860394243726933e-06,
+      "loss": 1.7177,
+      "step": 1213
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1846971870641542e-06,
+      "loss": 1.6807,
+      "step": 1214
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1833546051643324e-06,
+      "loss": 1.6868,
+      "step": 1215
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1820116811780948e-06,
+      "loss": 1.6993,
+      "step": 1216
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1806684176109484e-06,
+      "loss": 1.7035,
+      "step": 1217
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1793248169690318e-06,
+      "loss": 1.6899,
+      "step": 1218
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.177980881759114e-06,
+      "loss": 1.6628,
+      "step": 1219
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1766366144885876e-06,
+      "loss": 1.6772,
+      "step": 1220
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.175292017665464e-06,
+      "loss": 1.6627,
+      "step": 1221
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1739470937983706e-06,
+      "loss": 1.6871,
+      "step": 1222
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1726018453965449e-06,
+      "loss": 1.658,
+      "step": 1223
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.171256274969829e-06,
+      "loss": 1.673,
+      "step": 1224
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1699103850286667e-06,
+      "loss": 1.6777,
+      "step": 1225
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1685641780840972e-06,
+      "loss": 1.6653,
+      "step": 1226
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1672176566477518e-06,
+      "loss": 1.6911,
+      "step": 1227
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1658708232318482e-06,
+      "loss": 1.7045,
+      "step": 1228
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.164523680349186e-06,
+      "loss": 1.6691,
+      "step": 1229
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1631762305131422e-06,
+      "loss": 1.6937,
+      "step": 1230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1618284762376673e-06,
+      "loss": 1.6857,
+      "step": 1231
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1604804200372784e-06,
+      "loss": 1.6885,
+      "step": 1232
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1591320644270572e-06,
+      "loss": 1.6968,
+      "step": 1233
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1577834119226438e-06,
+      "loss": 1.6895,
+      "step": 1234
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.156434465040231e-06,
+      "loss": 1.6709,
+      "step": 1235
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.155085226296562e-06,
+      "loss": 1.6825,
+      "step": 1236
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1537356982089247e-06,
+      "loss": 1.6898,
+      "step": 1237
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1523858832951458e-06,
+      "loss": 1.657,
+      "step": 1238
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1510357840735882e-06,
+      "loss": 1.7059,
+      "step": 1239
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1496854030631443e-06,
+      "loss": 1.6842,
+      "step": 1240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1483347427832328e-06,
+      "loss": 1.6689,
+      "step": 1241
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1469838057537939e-06,
+      "loss": 1.6703,
+      "step": 1242
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1456325944952826e-06,
+      "loss": 1.6478,
+      "step": 1243
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1442811115286667e-06,
+      "loss": 1.6915,
+      "step": 1244
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1429293593754215e-06,
+      "loss": 1.6843,
+      "step": 1245
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.141577340557523e-06,
+      "loss": 1.6685,
+      "step": 1246
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1402250575974454e-06,
+      "loss": 1.6758,
+      "step": 1247
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1388725130181565e-06,
+      "loss": 1.6811,
+      "step": 1248
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1375197093431105e-06,
+      "loss": 1.6874,
+      "step": 1249
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1361666490962467e-06,
+      "loss": 1.6721,
+      "step": 1250
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.134813334801982e-06,
+      "loss": 1.6702,
+      "step": 1251
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1334597689852075e-06,
+      "loss": 1.6978,
+      "step": 1252
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1321059541712844e-06,
+      "loss": 1.684,
+      "step": 1253
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.130751892886037e-06,
+      "loss": 1.6696,
+      "step": 1254
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1293975876557504e-06,
+      "loss": 1.6795,
+      "step": 1255
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.128043041007165e-06,
+      "loss": 1.6947,
+      "step": 1256
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1266882554674707e-06,
+      "loss": 1.6825,
+      "step": 1257
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1253332335643042e-06,
+      "loss": 1.6716,
+      "step": 1258
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1239779778257424e-06,
+      "loss": 1.701,
+      "step": 1259
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1226224907802983e-06,
+      "loss": 1.6634,
+      "step": 1260
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1212667749569178e-06,
+      "loss": 1.6802,
+      "step": 1261
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1199108328849717e-06,
+      "loss": 1.6618,
+      "step": 1262
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1185546670942546e-06,
+      "loss": 1.6962,
+      "step": 1263
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1171982801149773e-06,
+      "loss": 1.6803,
+      "step": 1264
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1158416744777644e-06,
+      "loss": 1.6687,
+      "step": 1265
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.114484852713647e-06,
+      "loss": 1.6932,
+      "step": 1266
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1131278173540608e-06,
+      "loss": 1.6685,
+      "step": 1267
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1117705709308393e-06,
+      "loss": 1.6753,
+      "step": 1268
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1104131159762102e-06,
+      "loss": 1.688,
+      "step": 1269
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1090554550227898e-06,
+      "loss": 1.7025,
+      "step": 1270
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1076975906035789e-06,
+      "loss": 1.6913,
+      "step": 1271
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1063395252519578e-06,
+      "loss": 1.7012,
+      "step": 1272
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1049812615016823e-06,
+      "loss": 1.6939,
+      "step": 1273
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1036228018868773e-06,
+      "loss": 1.6564,
+      "step": 1274
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.102264148942034e-06,
+      "loss": 1.6598,
+      "step": 1275
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1009053052020044e-06,
+      "loss": 1.7008,
+      "step": 1276
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0995462732019956e-06,
+      "loss": 1.6683,
+      "step": 1277
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0981870554775664e-06,
+      "loss": 1.6517,
+      "step": 1278
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0968276545646222e-06,
+      "loss": 1.6829,
+      "step": 1279
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.09546807299941e-06,
+      "loss": 1.6604,
+      "step": 1280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0941083133185145e-06,
+      "loss": 1.7067,
+      "step": 1281
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0927483780588513e-06,
+      "loss": 1.6734,
+      "step": 1282
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0913882697576648e-06,
+      "loss": 1.6957,
+      "step": 1283
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0900279909525225e-06,
+      "loss": 1.687,
+      "step": 1284
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0886675441813083e-06,
+      "loss": 1.683,
+      "step": 1285
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.087306931982221e-06,
+      "loss": 1.653,
+      "step": 1286
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.085946156893768e-06,
+      "loss": 1.6623,
+      "step": 1287
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.08458522145476e-06,
+      "loss": 1.6853,
+      "step": 1288
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0832241282043066e-06,
+      "loss": 1.6421,
+      "step": 1289
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0818628796818131e-06,
+      "loss": 1.6546,
+      "step": 1290
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0805014784269734e-06,
+      "loss": 1.6969,
+      "step": 1291
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.079139926979766e-06,
+      "loss": 1.6692,
+      "step": 1292
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0777782278804511e-06,
+      "loss": 1.6688,
+      "step": 1293
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.076416383669563e-06,
+      "loss": 1.6668,
+      "step": 1294
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.075054396887908e-06,
+      "loss": 1.6527,
+      "step": 1295
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0736922700765569e-06,
+      "loss": 1.6502,
+      "step": 1296
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.072330005776843e-06,
+      "loss": 1.6875,
+      "step": 1297
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0709676065303555e-06,
+      "loss": 1.6897,
+      "step": 1298
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0696050748789357e-06,
+      "loss": 1.6878,
+      "step": 1299
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.068242413364671e-06,
+      "loss": 1.6944,
+      "step": 1300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0668796245298927e-06,
+      "loss": 1.6669,
+      "step": 1301
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0655167109171684e-06,
+      "loss": 1.6762,
+      "step": 1302
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0641536750692993e-06,
+      "loss": 1.6434,
+      "step": 1303
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0627905195293135e-06,
+      "loss": 1.6856,
+      "step": 1304
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0614272468404634e-06,
+      "loss": 1.6573,
+      "step": 1305
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.06006385954622e-06,
+      "loss": 1.6691,
+      "step": 1306
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.058700360190268e-06,
+      "loss": 1.6614,
+      "step": 1307
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0573367513165e-06,
+      "loss": 1.6697,
+      "step": 1308
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.055973035469015e-06,
+      "loss": 1.6758,
+      "step": 1309
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.05460921519211e-06,
+      "loss": 1.6609,
+      "step": 1310
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0532452930302776e-06,
+      "loss": 1.6667,
+      "step": 1311
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0518812715282e-06,
+      "loss": 1.6671,
+      "step": 1312
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0505171532307447e-06,
+      "loss": 1.6827,
+      "step": 1313
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0491529406829608e-06,
+      "loss": 1.6577,
+      "step": 1314
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.047788636430072e-06,
+      "loss": 1.6514,
+      "step": 1315
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0464242430174737e-06,
+      "loss": 1.6722,
+      "step": 1316
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0450597629907273e-06,
+      "loss": 1.6771,
+      "step": 1317
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.043695198895557e-06,
+      "loss": 1.7177,
+      "step": 1318
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0423305532778418e-06,
+      "loss": 1.6889,
+      "step": 1319
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0409658286836142e-06,
+      "loss": 1.6761,
+      "step": 1320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.039601027659054e-06,
+      "loss": 1.6992,
+      "step": 1321
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0382361527504834e-06,
+      "loss": 1.6904,
+      "step": 1322
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0368712065043619e-06,
+      "loss": 1.6702,
+      "step": 1323
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0355061914672829e-06,
+      "loss": 1.6715,
+      "step": 1324
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0341411101859678e-06,
+      "loss": 1.7106,
+      "step": 1325
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0327759652072616e-06,
+      "loss": 1.6946,
+      "step": 1326
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0314107590781281e-06,
+      "loss": 1.6842,
+      "step": 1327
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0300454943456455e-06,
+      "loss": 1.6499,
+      "step": 1328
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0286801735570008e-06,
+      "loss": 1.6649,
+      "step": 1329
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0273147992594859e-06,
+      "loss": 1.6558,
+      "step": 1330
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0259493740004925e-06,
+      "loss": 1.6833,
+      "step": 1331
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0245839003275073e-06,
+      "loss": 1.6699,
+      "step": 1332
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0232183807881078e-06,
+      "loss": 1.6553,
+      "step": 1333
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0218528179299562e-06,
+      "loss": 1.6453,
+      "step": 1334
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0204872143007962e-06,
+      "loss": 1.6737,
+      "step": 1335
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0191215724484474e-06,
+      "loss": 1.672,
+      "step": 1336
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0177558949208007e-06,
+      "loss": 1.7187,
+      "step": 1337
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0163901842658134e-06,
+      "loss": 1.6919,
+      "step": 1338
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0150244430315046e-06,
+      "loss": 1.688,
+      "step": 1339
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.013658673765951e-06,
+      "loss": 1.7159,
+      "step": 1340
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0122928790172813e-06,
+      "loss": 1.6773,
+      "step": 1341
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.010927061333671e-06,
+      "loss": 1.688,
+      "step": 1342
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0095612232633392e-06,
+      "loss": 1.6629,
+      "step": 1343
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0081953673545432e-06,
+      "loss": 1.6618,
+      "step": 1344
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0068294961555732e-06,
+      "loss": 1.687,
+      "step": 1345
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.005463612214748e-06,
+      "loss": 1.6546,
+      "step": 1346
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0040977180804095e-06,
+      "loss": 1.6803,
+      "step": 1347
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.00273181630092e-06,
+      "loss": 1.6566,
+      "step": 1348
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.001365909424655e-06,
+      "loss": 1.6478,
+      "step": 1349
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1e-06,
+      "loss": 1.6594,
+      "step": 1350
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.986340905753449e-07,
+      "loss": 1.6922,
+      "step": 1351
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.972681836990799e-07,
+      "loss": 1.6454,
+      "step": 1352
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.959022819195908e-07,
+      "loss": 1.6857,
+      "step": 1353
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.945363877852524e-07,
+      "loss": 1.6601,
+      "step": 1354
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.931705038444267e-07,
+      "loss": 1.6841,
+      "step": 1355
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.918046326454567e-07,
+      "loss": 1.6781,
+      "step": 1356
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.904387767366607e-07,
+      "loss": 1.6617,
+      "step": 1357
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.89072938666329e-07,
+      "loss": 1.685,
+      "step": 1358
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.87707120982719e-07,
+      "loss": 1.6821,
+      "step": 1359
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.863413262340491e-07,
+      "loss": 1.6559,
+      "step": 1360
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.849755569684953e-07,
+      "loss": 1.6612,
+      "step": 1361
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.836098157341867e-07,
+      "loss": 1.6767,
+      "step": 1362
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.822441050791994e-07,
+      "loss": 1.6621,
+      "step": 1363
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.808784275515525e-07,
+      "loss": 1.6526,
+      "step": 1364
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.795127856992039e-07,
+      "loss": 1.6732,
+      "step": 1365
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.78147182070044e-07,
+      "loss": 1.7162,
+      "step": 1366
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.767816192118921e-07,
+      "loss": 1.7013,
+      "step": 1367
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.754160996724926e-07,
+      "loss": 1.6923,
+      "step": 1368
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.740506259995074e-07,
+      "loss": 1.6799,
+      "step": 1369
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.726852007405142e-07,
+      "loss": 1.682,
+      "step": 1370
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.713198264429993e-07,
+      "loss": 1.6579,
+      "step": 1371
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.699545056543546e-07,
+      "loss": 1.6714,
+      "step": 1372
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.685892409218718e-07,
+      "loss": 1.6518,
+      "step": 1373
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.67224034792738e-07,
+      "loss": 1.6666,
+      "step": 1374
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.658588898140321e-07,
+      "loss": 1.6681,
+      "step": 1375
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.644938085327173e-07,
+      "loss": 1.7044,
+      "step": 1376
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.63128793495638e-07,
+      "loss": 1.6691,
+      "step": 1377
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.617638472495167e-07,
+      "loss": 1.6669,
+      "step": 1378
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.603989723409458e-07,
+      "loss": 1.6715,
+      "step": 1379
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.590341713163857e-07,
+      "loss": 1.6736,
+      "step": 1380
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.576694467221581e-07,
+      "loss": 1.6738,
+      "step": 1381
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.563048011044432e-07,
+      "loss": 1.6651,
+      "step": 1382
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.549402370092726e-07,
+      "loss": 1.6756,
+      "step": 1383
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.535757569825265e-07,
+      "loss": 1.6797,
+      "step": 1384
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.52211363569928e-07,
+      "loss": 1.6655,
+      "step": 1385
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.508470593170392e-07,
+      "loss": 1.6644,
+      "step": 1386
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.494828467692556e-07,
+      "loss": 1.6889,
+      "step": 1387
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.481187284718005e-07,
+      "loss": 1.6695,
+      "step": 1388
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.467547069697226e-07,
+      "loss": 1.6843,
+      "step": 1389
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.453907848078901e-07,
+      "loss": 1.666,
+      "step": 1390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.44026964530985e-07,
+      "loss": 1.6526,
+      "step": 1391
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.426632486834997e-07,
+      "loss": 1.6772,
+      "step": 1392
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.412996398097323e-07,
+      "loss": 1.6807,
+      "step": 1393
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.399361404537799e-07,
+      "loss": 1.6843,
+      "step": 1394
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.385727531595366e-07,
+      "loss": 1.7006,
+      "step": 1395
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.372094804706866e-07,
+      "loss": 1.6721,
+      "step": 1396
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.358463249307008e-07,
+      "loss": 1.6778,
+      "step": 1397
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.344832890828316e-07,
+      "loss": 1.7063,
+      "step": 1398
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.331203754701074e-07,
+      "loss": 1.6659,
+      "step": 1399
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.317575866353291e-07,
+      "loss": 1.7271,
+      "step": 1400
+    },
+    {
+      "epoch": 0.98,
+      "eval_loss": 1.6688400506973267,
+      "eval_runtime": 69.6998,
+      "eval_samples_per_second": 28.695,
+      "eval_steps_per_second": 0.072,
+      "step": 1400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.303949251210645e-07,
+      "loss": 1.6517,
+      "step": 1401
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.290323934696446e-07,
+      "loss": 1.6941,
+      "step": 1402
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.27669994223157e-07,
+      "loss": 1.6625,
+      "step": 1403
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.263077299234432e-07,
+      "loss": 1.7042,
+      "step": 1404
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.249456031120922e-07,
+      "loss": 1.691,
+      "step": 1405
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.235836163304369e-07,
+      "loss": 1.6892,
+      "step": 1406
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.22221772119549e-07,
+      "loss": 1.6769,
+      "step": 1407
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.208600730202339e-07,
+      "loss": 1.6724,
+      "step": 1408
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.194985215730269e-07,
+      "loss": 1.6732,
+      "step": 1409
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.181371203181871e-07,
+      "loss": 1.6992,
+      "step": 1410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.167758717956933e-07,
+      "loss": 1.6851,
+      "step": 1411
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.1541477854524e-07,
+      "loss": 1.6969,
+      "step": 1412
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.140538431062319e-07,
+      "loss": 1.6594,
+      "step": 1413
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.126930680177787e-07,
+      "loss": 1.6724,
+      "step": 1414
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.11332455818692e-07,
+      "loss": 1.6805,
+      "step": 1415
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.099720090474779e-07,
+      "loss": 1.6838,
+      "step": 1416
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.086117302423351e-07,
+      "loss": 1.6684,
+      "step": 1417
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.072516219411488e-07,
+      "loss": 1.658,
+      "step": 1418
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.058916866814856e-07,
+      "loss": 1.6772,
+      "step": 1419
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.045319270005899e-07,
+      "loss": 1.6953,
+      "step": 1420
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.031723454353779e-07,
+      "loss": 1.688,
+      "step": 1421
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.018129445224339e-07,
+      "loss": 1.6598,
+      "step": 1422
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.004537267980044e-07,
+      "loss": 1.6976,
+      "step": 1423
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.990946947979954e-07,
+      "loss": 1.6723,
+      "step": 1424
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.977358510579657e-07,
+      "loss": 1.6529,
+      "step": 1425
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.963771981131225e-07,
+      "loss": 1.648,
+      "step": 1426
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.950187384983179e-07,
+      "loss": 1.6781,
+      "step": 1427
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.936604747480422e-07,
+      "loss": 1.6266,
+      "step": 1428
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.923024093964213e-07,
+      "loss": 1.6759,
+      "step": 1429
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.909445449772102e-07,
+      "loss": 1.6768,
+      "step": 1430
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.895868840237895e-07,
+      "loss": 1.6604,
+      "step": 1431
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.882294290691607e-07,
+      "loss": 1.6955,
+      "step": 1432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.868721826459394e-07,
+      "loss": 1.6816,
+      "step": 1433
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.855151472863532e-07,
+      "loss": 1.6692,
+      "step": 1434
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.841583255222358e-07,
+      "loss": 1.6526,
+      "step": 1435
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.828017198850226e-07,
+      "loss": 1.6588,
+      "step": 1436
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.814453329057453e-07,
+      "loss": 1.6385,
+      "step": 1437
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.800891671150285e-07,
+      "loss": 1.6652,
+      "step": 1438
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.787332250430823e-07,
+      "loss": 1.65,
+      "step": 1439
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.773775092197017e-07,
+      "loss": 1.6555,
+      "step": 1440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.760220221742577e-07,
+      "loss": 1.6737,
+      "step": 1441
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.746667664356955e-07,
+      "loss": 1.6498,
+      "step": 1442
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.733117445325291e-07,
+      "loss": 1.6762,
+      "step": 1443
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.719569589928353e-07,
+      "loss": 1.6405,
+      "step": 1444
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.706024123442497e-07,
+      "loss": 1.6646,
+      "step": 1445
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.69248107113963e-07,
+      "loss": 1.6764,
+      "step": 1446
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.678940458287158e-07,
+      "loss": 1.6794,
+      "step": 1447
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.665402310147924e-07,
+      "loss": 1.6694,
+      "step": 1448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.651866651980183e-07,
+      "loss": 1.671,
+      "step": 1449
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.638333509037535e-07,
+      "loss": 1.67,
+      "step": 1450
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.624802906568895e-07,
+      "loss": 1.6518,
+      "step": 1451
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.611274869818437e-07,
+      "loss": 1.6698,
+      "step": 1452
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.597749424025544e-07,
+      "loss": 1.6587,
+      "step": 1453
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.58422659442477e-07,
+      "loss": 1.6288,
+      "step": 1454
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.570706406245786e-07,
+      "loss": 1.6757,
+      "step": 1455
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.557188884713332e-07,
+      "loss": 1.6555,
+      "step": 1456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.543674055047176e-07,
+      "loss": 1.6907,
+      "step": 1457
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.530161942462062e-07,
+      "loss": 1.6555,
+      "step": 1458
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.516652572167671e-07,
+      "loss": 1.6707,
+      "step": 1459
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.503145969368561e-07,
+      "loss": 1.654,
+      "step": 1460
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.48964215926412e-07,
+      "loss": 1.6681,
+      "step": 1461
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.476141167048541e-07,
+      "loss": 1.6258,
+      "step": 1462
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.462643017910755e-07,
+      "loss": 1.6663,
+      "step": 1463
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.449147737034379e-07,
+      "loss": 1.693,
+      "step": 1464
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.435655349597689e-07,
+      "loss": 1.6805,
+      "step": 1465
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.422165880773565e-07,
+      "loss": 1.6729,
+      "step": 1466
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.408679355729427e-07,
+      "loss": 1.6354,
+      "step": 1467
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.395195799627216e-07,
+      "loss": 1.6651,
+      "step": 1468
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.381715237623328e-07,
+      "loss": 1.6698,
+      "step": 1469
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.368237694868578e-07,
+      "loss": 1.6428,
+      "step": 1470
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.354763196508141e-07,
+      "loss": 1.6375,
+      "step": 1471
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.341291767681522e-07,
+      "loss": 1.6698,
+      "step": 1472
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.327823433522483e-07,
+      "loss": 1.6939,
+      "step": 1473
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.314358219159029e-07,
+      "loss": 1.661,
+      "step": 1474
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.300896149713334e-07,
+      "loss": 1.6772,
+      "step": 1475
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.287437250301708e-07,
+      "loss": 1.6623,
+      "step": 1476
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.273981546034552e-07,
+      "loss": 1.6872,
+      "step": 1477
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.260529062016293e-07,
+      "loss": 1.6839,
+      "step": 1478
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.247079823345362e-07,
+      "loss": 1.6983,
+      "step": 1479
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.233633855114126e-07,
+      "loss": 1.65,
+      "step": 1480
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.22019118240886e-07,
+      "loss": 1.6827,
+      "step": 1481
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.20675183030968e-07,
+      "loss": 1.6846,
+      "step": 1482
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.193315823890519e-07,
+      "loss": 1.6707,
+      "step": 1483
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.179883188219051e-07,
+      "loss": 1.6649,
+      "step": 1484
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.166453948356677e-07,
+      "loss": 1.6382,
+      "step": 1485
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.153028129358457e-07,
+      "loss": 1.6631,
+      "step": 1486
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.139605756273066e-07,
+      "loss": 1.6563,
+      "step": 1487
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.126186854142751e-07,
+      "loss": 1.6382,
+      "step": 1488
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.112771448003292e-07,
+      "loss": 1.6676,
+      "step": 1489
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.09935956288393e-07,
+      "loss": 1.6406,
+      "step": 1490
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.085951223807344e-07,
+      "loss": 1.6823,
+      "step": 1491
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.072546455789604e-07,
+      "loss": 1.6753,
+      "step": 1492
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.059145283840113e-07,
+      "loss": 1.6572,
+      "step": 1493
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.045747732961563e-07,
+      "loss": 1.6759,
+      "step": 1494
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.032353828149888e-07,
+      "loss": 1.6592,
+      "step": 1495
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.01896359439422e-07,
+      "loss": 1.6739,
+      "step": 1496
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.005577056676854e-07,
+      "loss": 1.6525,
+      "step": 1497
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.992194239973169e-07,
+      "loss": 1.6552,
+      "step": 1498
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.978815169251618e-07,
+      "loss": 1.688,
+      "step": 1499
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.965439869473663e-07,
+      "loss": 1.6626,
+      "step": 1500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.95206836559372e-07,
+      "loss": 1.6731,
+      "step": 1501
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.938700682559132e-07,
+      "loss": 1.6497,
+      "step": 1502
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.92533684531011e-07,
+      "loss": 1.6578,
+      "step": 1503
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.911976878779695e-07,
+      "loss": 1.6606,
+      "step": 1504
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.898620807893697e-07,
+      "loss": 1.6696,
+      "step": 1505
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.885268657570672e-07,
+      "loss": 1.6381,
+      "step": 1506
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.871920452721843e-07,
+      "loss": 1.6349,
+      "step": 1507
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.858576218251088e-07,
+      "loss": 1.6866,
+      "step": 1508
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.845235979054867e-07,
+      "loss": 1.683,
+      "step": 1509
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.831899760022192e-07,
+      "loss": 1.6681,
+      "step": 1510
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.818567586034576e-07,
+      "loss": 1.6462,
+      "step": 1511
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.805239481965975e-07,
+      "loss": 1.6653,
+      "step": 1512
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.791915472682762e-07,
+      "loss": 1.6434,
+      "step": 1513
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.778595583043665e-07,
+      "loss": 1.636,
+      "step": 1514
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.76527983789973e-07,
+      "loss": 1.6462,
+      "step": 1515
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.751968262094268e-07,
+      "loss": 1.6772,
+      "step": 1516
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.738660880462813e-07,
+      "loss": 1.6666,
+      "step": 1517
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.725357717833066e-07,
+      "loss": 1.6923,
+      "step": 1518
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.712058799024868e-07,
+      "loss": 1.6552,
+      "step": 1519
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.698764148850136e-07,
+      "loss": 1.6774,
+      "step": 1520
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.685473792112824e-07,
+      "loss": 1.6211,
+      "step": 1521
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.67218775360888e-07,
+      "loss": 1.725,
+      "step": 1522
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.658906058126182e-07,
+      "loss": 1.678,
+      "step": 1523
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.645628730444523e-07,
+      "loss": 1.6658,
+      "step": 1524
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.632355795335532e-07,
+      "loss": 1.6685,
+      "step": 1525
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.619087277562655e-07,
+      "loss": 1.6495,
+      "step": 1526
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.605823201881088e-07,
+      "loss": 1.6482,
+      "step": 1527
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.592563593037746e-07,
+      "loss": 1.6838,
+      "step": 1528
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.579308475771195e-07,
+      "loss": 1.666,
+      "step": 1529
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.566057874811642e-07,
+      "loss": 1.6742,
+      "step": 1530
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.552811814880857e-07,
+      "loss": 1.6447,
+      "step": 1531
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.539570320692137e-07,
+      "loss": 1.6573,
+      "step": 1532
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.526333416950261e-07,
+      "loss": 1.7023,
+      "step": 1533
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.513101128351453e-07,
+      "loss": 1.6685,
+      "step": 1534
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.49987347958331e-07,
+      "loss": 1.6611,
+      "step": 1535
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.486650495324782e-07,
+      "loss": 1.6607,
+      "step": 1536
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.473432200246117e-07,
+      "loss": 1.677,
+      "step": 1537
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.460218619008816e-07,
+      "loss": 1.6601,
+      "step": 1538
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.447009776265577e-07,
+      "loss": 1.648,
+      "step": 1539
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.433805696660265e-07,
+      "loss": 1.6586,
+      "step": 1540
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.420606404827854e-07,
+      "loss": 1.6795,
+      "step": 1541
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.407411925394387e-07,
+      "loss": 1.6262,
+      "step": 1542
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.394222282976935e-07,
+      "loss": 1.6843,
+      "step": 1543
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.381037502183527e-07,
+      "loss": 1.6542,
+      "step": 1544
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.367857607613146e-07,
+      "loss": 1.6638,
+      "step": 1545
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.354682623855633e-07,
+      "loss": 1.6576,
+      "step": 1546
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.341512575491688e-07,
+      "loss": 1.6595,
+      "step": 1547
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.32834748709279e-07,
+      "loss": 1.6678,
+      "step": 1548
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.315187383221169e-07,
+      "loss": 1.6848,
+      "step": 1549
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.302032288429756e-07,
+      "loss": 1.6413,
+      "step": 1550
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.288882227262139e-07,
+      "loss": 1.6547,
+      "step": 1551
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.275737224252502e-07,
+      "loss": 1.674,
+      "step": 1552
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.262597303925609e-07,
+      "loss": 1.6382,
+      "step": 1553
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.24946249079673e-07,
+      "loss": 1.6721,
+      "step": 1554
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.236332809371608e-07,
+      "loss": 1.6569,
+      "step": 1555
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.22320828414642e-07,
+      "loss": 1.667,
+      "step": 1556
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.210088939607708e-07,
+      "loss": 1.6817,
+      "step": 1557
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.196974800232363e-07,
+      "loss": 1.6435,
+      "step": 1558
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.183865890487554e-07,
+      "loss": 1.6495,
+      "step": 1559
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.170762234830698e-07,
+      "loss": 1.6658,
+      "step": 1560
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.157663857709415e-07,
+      "loss": 1.6762,
+      "step": 1561
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.144570783561468e-07,
+      "loss": 1.6634,
+      "step": 1562
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.131483036814721e-07,
+      "loss": 1.6457,
+      "step": 1563
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.118400641887115e-07,
+      "loss": 1.6617,
+      "step": 1564
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.105323623186594e-07,
+      "loss": 1.6643,
+      "step": 1565
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.092252005111077e-07,
+      "loss": 1.6627,
+      "step": 1566
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.079185812048403e-07,
+      "loss": 1.6859,
+      "step": 1567
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.066125068376296e-07,
+      "loss": 1.6717,
+      "step": 1568
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.053069798462302e-07,
+      "loss": 1.6851,
+      "step": 1569
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.040020026663766e-07,
+      "loss": 1.6698,
+      "step": 1570
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.026975777327768e-07,
+      "loss": 1.6488,
+      "step": 1571
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.013937074791087e-07,
+      "loss": 1.6626,
+      "step": 1572
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.000903943380159e-07,
+      "loss": 1.6549,
+      "step": 1573
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.98787640741101e-07,
+      "loss": 1.6678,
+      "step": 1574
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.974854491189243e-07,
+      "loss": 1.6666,
+      "step": 1575
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.961838219009966e-07,
+      "loss": 1.6964,
+      "step": 1576
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.948827615157767e-07,
+      "loss": 1.6813,
+      "step": 1577
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.935822703906646e-07,
+      "loss": 1.6571,
+      "step": 1578
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.922823509519994e-07,
+      "loss": 1.6728,
+      "step": 1579
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.909830056250526e-07,
+      "loss": 1.6926,
+      "step": 1580
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.896842368340252e-07,
+      "loss": 1.6559,
+      "step": 1581
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.883860470020421e-07,
+      "loss": 1.6619,
+      "step": 1582
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.870884385511488e-07,
+      "loss": 1.6646,
+      "step": 1583
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.857914139023057e-07,
+      "loss": 1.6873,
+      "step": 1584
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.844949754753832e-07,
+      "loss": 1.6799,
+      "step": 1585
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.831991256891591e-07,
+      "loss": 1.6789,
+      "step": 1586
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.819038669613124e-07,
+      "loss": 1.6549,
+      "step": 1587
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.806092017084199e-07,
+      "loss": 1.643,
+      "step": 1588
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.793151323459505e-07,
+      "loss": 1.6692,
+      "step": 1589
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.780216612882619e-07,
+      "loss": 1.6729,
+      "step": 1590
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.767287909485946e-07,
+      "loss": 1.6769,
+      "step": 1591
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.754365237390696e-07,
+      "loss": 1.6822,
+      "step": 1592
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.741448620706815e-07,
+      "loss": 1.6654,
+      "step": 1593
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.72853808353296e-07,
+      "loss": 1.6962,
+      "step": 1594
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.715633649956444e-07,
+      "loss": 1.6409,
+      "step": 1595
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.702735344053186e-07,
+      "loss": 1.656,
+      "step": 1596
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.68984318988768e-07,
+      "loss": 1.6554,
+      "step": 1597
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.676957211512934e-07,
+      "loss": 1.683,
+      "step": 1598
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.664077432970445e-07,
+      "loss": 1.6752,
+      "step": 1599
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.651203878290138e-07,
+      "loss": 1.6578,
+      "step": 1600
+    },
+    {
+      "epoch": 1.12,
+      "eval_loss": 1.6659311056137085,
+      "eval_runtime": 71.8369,
+      "eval_samples_per_second": 27.841,
+      "eval_steps_per_second": 0.07,
+      "step": 1600
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.638336571490324e-07,
+      "loss": 1.6529,
+      "step": 1601
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.625475536577654e-07,
+      "loss": 1.6554,
+      "step": 1602
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.612620797547086e-07,
+      "loss": 1.6705,
+      "step": 1603
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.59977237838183e-07,
+      "loss": 1.6587,
+      "step": 1604
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.586930303053296e-07,
+      "loss": 1.6879,
+      "step": 1605
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.574094595521071e-07,
+      "loss": 1.6539,
+      "step": 1606
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.561265279732857e-07,
+      "loss": 1.666,
+      "step": 1607
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.548442379624424e-07,
+      "loss": 1.6577,
+      "step": 1608
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.535625919119578e-07,
+      "loss": 1.6513,
+      "step": 1609
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.522815922130111e-07,
+      "loss": 1.6501,
+      "step": 1610
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.510012412555758e-07,
+      "loss": 1.6713,
+      "step": 1611
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.497215414284145e-07,
+      "loss": 1.6782,
+      "step": 1612
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.484424951190757e-07,
+      "loss": 1.6533,
+      "step": 1613
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.471641047138872e-07,
+      "loss": 1.6808,
+      "step": 1614
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.458863725979549e-07,
+      "loss": 1.6795,
+      "step": 1615
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.44609301155155e-07,
+      "loss": 1.658,
+      "step": 1616
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.433328927681322e-07,
+      "loss": 1.642,
+      "step": 1617
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.420571498182938e-07,
+      "loss": 1.651,
+      "step": 1618
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.40782074685805e-07,
+      "loss": 1.6741,
+      "step": 1619
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.395076697495854e-07,
+      "loss": 1.6381,
+      "step": 1620
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.382339373873046e-07,
+      "loss": 1.6989,
+      "step": 1621
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.369608799753771e-07,
+      "loss": 1.648,
+      "step": 1622
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.356884998889581e-07,
+      "loss": 1.6719,
+      "step": 1623
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.344167995019394e-07,
+      "loss": 1.6507,
+      "step": 1624
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.331457811869436e-07,
+      "loss": 1.6613,
+      "step": 1625
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.31875447315322e-07,
+      "loss": 1.6396,
+      "step": 1626
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.306058002571487e-07,
+      "loss": 1.6881,
+      "step": 1627
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.293368423812158e-07,
+      "loss": 1.6684,
+      "step": 1628
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.280685760550302e-07,
+      "loss": 1.6522,
+      "step": 1629
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.268010036448087e-07,
+      "loss": 1.6728,
+      "step": 1630
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.255341275154724e-07,
+      "loss": 1.6697,
+      "step": 1631
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.242679500306443e-07,
+      "loss": 1.6799,
+      "step": 1632
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.230024735526436e-07,
+      "loss": 1.6841,
+      "step": 1633
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.217377004424818e-07,
+      "loss": 1.6351,
+      "step": 1634
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.204736330598584e-07,
+      "loss": 1.6575,
+      "step": 1635
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.19210273763155e-07,
+      "loss": 1.6494,
+      "step": 1636
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.179476249094336e-07,
+      "loss": 1.6475,
+      "step": 1637
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.166856888544296e-07,
+      "loss": 1.6594,
+      "step": 1638
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.154244679525493e-07,
+      "loss": 1.6381,
+      "step": 1639
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.141639645568645e-07,
+      "loss": 1.6772,
+      "step": 1640
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.129041810191083e-07,
+      "loss": 1.6559,
+      "step": 1641
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.116451196896706e-07,
+      "loss": 1.6668,
+      "step": 1642
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.103867829175935e-07,
+      "loss": 1.6826,
+      "step": 1643
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.091291730505684e-07,
+      "loss": 1.6605,
+      "step": 1644
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.078722924349299e-07,
+      "loss": 1.6699,
+      "step": 1645
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.06616143415652e-07,
+      "loss": 1.648,
+      "step": 1646
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.053607283363431e-07,
+      "loss": 1.6587,
+      "step": 1647
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.041060495392436e-07,
+      "loss": 1.6733,
+      "step": 1648
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.028521093652194e-07,
+      "loss": 1.6765,
+      "step": 1649
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.015989101537586e-07,
+      "loss": 1.6564,
+      "step": 1650
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.003464542429665e-07,
+      "loss": 1.6474,
+      "step": 1651
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.990947439695624e-07,
+      "loss": 1.6518,
+      "step": 1652
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.978437816688733e-07,
+      "loss": 1.6813,
+      "step": 1653
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.965935696748321e-07,
+      "loss": 1.6707,
+      "step": 1654
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.953441103199704e-07,
+      "loss": 1.6572,
+      "step": 1655
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.940954059354164e-07,
+      "loss": 1.6662,
+      "step": 1656
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.928474588508903e-07,
+      "loss": 1.6486,
+      "step": 1657
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.916002713946983e-07,
+      "loss": 1.6831,
+      "step": 1658
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.903538458937294e-07,
+      "loss": 1.6523,
+      "step": 1659
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.891081846734518e-07,
+      "loss": 1.6636,
+      "step": 1660
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.878632900579076e-07,
+      "loss": 1.6816,
+      "step": 1661
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.866191643697078e-07,
+      "loss": 1.662,
+      "step": 1662
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.853758099300305e-07,
+      "loss": 1.6687,
+      "step": 1663
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.841332290586126e-07,
+      "loss": 1.6572,
+      "step": 1664
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.828914240737494e-07,
+      "loss": 1.6529,
+      "step": 1665
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.816503972922884e-07,
+      "loss": 1.6367,
+      "step": 1666
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.804101510296244e-07,
+      "loss": 1.6834,
+      "step": 1667
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.791706875996974e-07,
+      "loss": 1.632,
+      "step": 1668
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.779320093149855e-07,
+      "loss": 1.6609,
+      "step": 1669
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.766941184865023e-07,
+      "loss": 1.6777,
+      "step": 1670
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.754570174237927e-07,
+      "loss": 1.647,
+      "step": 1671
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.742207084349273e-07,
+      "loss": 1.6713,
+      "step": 1672
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.729851938265002e-07,
+      "loss": 1.6541,
+      "step": 1673
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.717504759036222e-07,
+      "loss": 1.6574,
+      "step": 1674
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.705165569699182e-07,
+      "loss": 1.6507,
+      "step": 1675
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.692834393275225e-07,
+      "loss": 1.6796,
+      "step": 1676
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.680511252770739e-07,
+      "loss": 1.6609,
+      "step": 1677
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.668196171177128e-07,
+      "loss": 1.6642,
+      "step": 1678
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.655889171470758e-07,
+      "loss": 1.6689,
+      "step": 1679
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.643590276612909e-07,
+      "loss": 1.6706,
+      "step": 1680
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.631299509549748e-07,
+      "loss": 1.654,
+      "step": 1681
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.619016893212272e-07,
+      "loss": 1.6671,
+      "step": 1682
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.606742450516274e-07,
+      "loss": 1.6728,
+      "step": 1683
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.594476204362303e-07,
+      "loss": 1.6557,
+      "step": 1684
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.582218177635606e-07,
+      "loss": 1.6985,
+      "step": 1685
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.569968393206099e-07,
+      "loss": 1.6435,
+      "step": 1686
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.557726873928322e-07,
+      "loss": 1.6442,
+      "step": 1687
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.545493642641388e-07,
+      "loss": 1.6733,
+      "step": 1688
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.533268722168959e-07,
+      "loss": 1.6655,
+      "step": 1689
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.521052135319182e-07,
+      "loss": 1.6493,
+      "step": 1690
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.508843904884652e-07,
+      "loss": 1.6451,
+      "step": 1691
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.496644053642393e-07,
+      "loss": 1.6489,
+      "step": 1692
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.484452604353771e-07,
+      "loss": 1.6746,
+      "step": 1693
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.472269579764486e-07,
+      "loss": 1.6533,
+      "step": 1694
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.460095002604532e-07,
+      "loss": 1.6582,
+      "step": 1695
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.447928895588127e-07,
+      "loss": 1.6621,
+      "step": 1696
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.435771281413693e-07,
+      "loss": 1.6438,
+      "step": 1697
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.423622182763804e-07,
+      "loss": 1.6727,
+      "step": 1698
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.411481622305144e-07,
+      "loss": 1.671,
+      "step": 1699
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.399349622688478e-07,
+      "loss": 1.6841,
+      "step": 1700
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.387226206548592e-07,
+      "loss": 1.647,
+      "step": 1701
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.37511139650425e-07,
+      "loss": 1.6868,
+      "step": 1702
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.363005215158179e-07,
+      "loss": 1.6375,
+      "step": 1703
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.350907685096983e-07,
+      "loss": 1.6579,
+      "step": 1704
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.338818828891147e-07,
+      "loss": 1.71,
+      "step": 1705
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.326738669094961e-07,
+      "loss": 1.6818,
+      "step": 1706
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.314667228246488e-07,
+      "loss": 1.6936,
+      "step": 1707
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.302604528867544e-07,
+      "loss": 1.6855,
+      "step": 1708
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.290550593463605e-07,
+      "loss": 1.6883,
+      "step": 1709
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.278505444523816e-07,
+      "loss": 1.6725,
+      "step": 1710
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.266469104520927e-07,
+      "loss": 1.6393,
+      "step": 1711
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.254441595911254e-07,
+      "loss": 1.6489,
+      "step": 1712
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.242422941134625e-07,
+      "loss": 1.6737,
+      "step": 1713
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.23041316261437e-07,
+      "loss": 1.6861,
+      "step": 1714
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.218412282757231e-07,
+      "loss": 1.6595,
+      "step": 1715
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.206420323953373e-07,
+      "loss": 1.6704,
+      "step": 1716
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.194437308576305e-07,
+      "loss": 1.6663,
+      "step": 1717
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.182463258982846e-07,
+      "loss": 1.6602,
+      "step": 1718
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.1704981975131e-07,
+      "loss": 1.6693,
+      "step": 1719
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.158542146490399e-07,
+      "loss": 1.6753,
+      "step": 1720
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.146595128221245e-07,
+      "loss": 1.6653,
+      "step": 1721
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.134657164995314e-07,
+      "loss": 1.652,
+      "step": 1722
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.122728279085375e-07,
+      "loss": 1.6999,
+      "step": 1723
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.110808492747258e-07,
+      "loss": 1.639,
+      "step": 1724
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.09889782821983e-07,
+      "loss": 1.6629,
+      "step": 1725
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.08699630772492e-07,
+      "loss": 1.6706,
+      "step": 1726
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.075103953467314e-07,
+      "loss": 1.6693,
+      "step": 1727
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.063220787634686e-07,
+      "loss": 1.6396,
+      "step": 1728
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.051346832397569e-07,
+      "loss": 1.6896,
+      "step": 1729
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.039482109909319e-07,
+      "loss": 1.6436,
+      "step": 1730
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.027626642306056e-07,
+      "loss": 1.6835,
+      "step": 1731
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.015780451706641e-07,
+      "loss": 1.662,
+      "step": 1732
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.00394356021262e-07,
+      "loss": 1.6424,
+      "step": 1733
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.992115989908191e-07,
+      "loss": 1.6613,
+      "step": 1734
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.98029776286017e-07,
+      "loss": 1.6743,
+      "step": 1735
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.968488901117934e-07,
+      "loss": 1.6867,
+      "step": 1736
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.956689426713383e-07,
+      "loss": 1.6653,
+      "step": 1737
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.944899361660909e-07,
+      "loss": 1.6492,
+      "step": 1738
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.933118727957349e-07,
+      "loss": 1.677,
+      "step": 1739
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.921347547581939e-07,
+      "loss": 1.6344,
+      "step": 1740
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.909585842496287e-07,
+      "loss": 1.6292,
+      "step": 1741
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.897833634644313e-07,
+      "loss": 1.6597,
+      "step": 1742
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.886090945952224e-07,
+      "loss": 1.6582,
+      "step": 1743
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.874357798328464e-07,
+      "loss": 1.6725,
+      "step": 1744
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.862634213663671e-07,
+      "loss": 1.6628,
+      "step": 1745
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.850920213830659e-07,
+      "loss": 1.6358,
+      "step": 1746
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.839215820684341e-07,
+      "loss": 1.6844,
+      "step": 1747
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.827521056061717e-07,
+      "loss": 1.6624,
+      "step": 1748
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.815835941781815e-07,
+      "loss": 1.6657,
+      "step": 1749
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.804160499645667e-07,
+      "loss": 1.6725,
+      "step": 1750
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.792494751436249e-07,
+      "loss": 1.6465,
+      "step": 1751
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.780838718918467e-07,
+      "loss": 1.6603,
+      "step": 1752
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.769192423839084e-07,
+      "loss": 1.6442,
+      "step": 1753
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7575558879267075e-07,
+      "loss": 1.6455,
+      "step": 1754
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.745929132891727e-07,
+      "loss": 1.6648,
+      "step": 1755
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.734312180426289e-07,
+      "loss": 1.6409,
+      "step": 1756
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7227050522042557e-07,
+      "loss": 1.655,
+      "step": 1757
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.711107769881153e-07,
+      "loss": 1.6843,
+      "step": 1758
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.699520355094142e-07,
+      "loss": 1.6332,
+      "step": 1759
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.687942829461968e-07,
+      "loss": 1.681,
+      "step": 1760
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.676375214584929e-07,
+      "loss": 1.6648,
+      "step": 1761
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.664817532044839e-07,
+      "loss": 1.6586,
+      "step": 1762
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.6532698034049724e-07,
+      "loss": 1.6397,
+      "step": 1763
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.641732050210031e-07,
+      "loss": 1.6472,
+      "step": 1764
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.6302042939861216e-07,
+      "loss": 1.656,
+      "step": 1765
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.618686556240675e-07,
+      "loss": 1.6755,
+      "step": 1766
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.607178858462445e-07,
+      "loss": 1.6884,
+      "step": 1767
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.595681222121458e-07,
+      "loss": 1.6372,
+      "step": 1768
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.5841936686689595e-07,
+      "loss": 1.6392,
+      "step": 1769
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.572716219537385e-07,
+      "loss": 1.6632,
+      "step": 1770
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.5612488961403206e-07,
+      "loss": 1.6739,
+      "step": 1771
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.5497917198724567e-07,
+      "loss": 1.6722,
+      "step": 1772
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.538344712109561e-07,
+      "loss": 1.6864,
+      "step": 1773
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.5269078942084204e-07,
+      "loss": 1.6593,
+      "step": 1774
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.51548128750681e-07,
+      "loss": 1.6542,
+      "step": 1775
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.504064913323471e-07,
+      "loss": 1.6731,
+      "step": 1776
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.492658792958026e-07,
+      "loss": 1.6676,
+      "step": 1777
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.481262947690982e-07,
+      "loss": 1.6548,
+      "step": 1778
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.4698773987836847e-07,
+      "loss": 1.658,
+      "step": 1779
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.458502167478253e-07,
+      "loss": 1.6811,
+      "step": 1780
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.447137274997562e-07,
+      "loss": 1.6417,
+      "step": 1781
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.435782742545209e-07,
+      "loss": 1.6615,
+      "step": 1782
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.424438591305437e-07,
+      "loss": 1.6749,
+      "step": 1783
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.4131048424431483e-07,
+      "loss": 1.6486,
+      "step": 1784
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.4017815171038184e-07,
+      "loss": 1.6652,
+      "step": 1785
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3904686364134826e-07,
+      "loss": 1.6269,
+      "step": 1786
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3791662214786963e-07,
+      "loss": 1.6642,
+      "step": 1787
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.36787429338647e-07,
+      "loss": 1.6698,
+      "step": 1788
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.356592873204269e-07,
+      "loss": 1.679,
+      "step": 1789
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3453219819799416e-07,
+      "loss": 1.67,
+      "step": 1790
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.334061640741694e-07,
+      "loss": 1.6675,
+      "step": 1791
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3228118704980577e-07,
+      "loss": 1.66,
+      "step": 1792
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3115726922378335e-07,
+      "loss": 1.64,
+      "step": 1793
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.300344126930053e-07,
+      "loss": 1.6825,
+      "step": 1794
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.289126195523968e-07,
+      "loss": 1.6547,
+      "step": 1795
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.2779189189489727e-07,
+      "loss": 1.6524,
+      "step": 1796
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.266722318114587e-07,
+      "loss": 1.689,
+      "step": 1797
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.255536413910421e-07,
+      "loss": 1.6763,
+      "step": 1798
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.244361227206118e-07,
+      "loss": 1.68,
+      "step": 1799
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.233196778851329e-07,
+      "loss": 1.6343,
+      "step": 1800
+    },
+    {
+      "epoch": 1.26,
+      "eval_loss": 1.6638457775115967,
+      "eval_runtime": 77.4679,
+      "eval_samples_per_second": 25.817,
+      "eval_steps_per_second": 0.065,
+      "step": 1800
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.222043089675672e-07,
+      "loss": 1.6697,
+      "step": 1801
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.2109001804886843e-07,
+      "loss": 1.6525,
+      "step": 1802
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.1997680720798057e-07,
+      "loss": 1.6991,
+      "step": 1803
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.188646785218315e-07,
+      "loss": 1.661,
+      "step": 1804
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.1775363406532905e-07,
+      "loss": 1.6981,
+      "step": 1805
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.1664367591136054e-07,
+      "loss": 1.6693,
+      "step": 1806
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.1553480613078483e-07,
+      "loss": 1.6464,
+      "step": 1807
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.1442702679243055e-07,
+      "loss": 1.6779,
+      "step": 1808
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.1332033996309267e-07,
+      "loss": 1.6669,
+      "step": 1809
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.1221474770752696e-07,
+      "loss": 1.6712,
+      "step": 1810
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.111102520884472e-07,
+      "loss": 1.6628,
+      "step": 1811
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.1000685516652136e-07,
+      "loss": 1.6446,
+      "step": 1812
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.08904559000367e-07,
+      "loss": 1.6502,
+      "step": 1813
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.0780336564654937e-07,
+      "loss": 1.677,
+      "step": 1814
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.0670327715957485e-07,
+      "loss": 1.66,
+      "step": 1815
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.0560429559188894e-07,
+      "loss": 1.6721,
+      "step": 1816
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.045064229938717e-07,
+      "loss": 1.6343,
+      "step": 1817
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.034096614138343e-07,
+      "loss": 1.6788,
+      "step": 1818
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.023140128980157e-07,
+      "loss": 1.6726,
+      "step": 1819
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.0121947949057745e-07,
+      "loss": 1.6562,
+      "step": 1820
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.0012606323360076e-07,
+      "loss": 1.6455,
+      "step": 1821
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.9903376616708265e-07,
+      "loss": 1.6521,
+      "step": 1822
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.9794259032893196e-07,
+      "loss": 1.6745,
+      "step": 1823
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.9685253775496566e-07,
+      "loss": 1.6307,
+      "step": 1824
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.957636104789055e-07,
+      "loss": 1.6454,
+      "step": 1825
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.946758105323732e-07,
+      "loss": 1.6547,
+      "step": 1826
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.935891399448873e-07,
+      "loss": 1.6364,
+      "step": 1827
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.9250360074385934e-07,
+      "loss": 1.659,
+      "step": 1828
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.9141919495458986e-07,
+      "loss": 1.6896,
+      "step": 1829
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.9033592460026543e-07,
+      "loss": 1.6335,
+      "step": 1830
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.8925379170195363e-07,
+      "loss": 1.6684,
+      "step": 1831
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.8817279827859983e-07,
+      "loss": 1.6775,
+      "step": 1832
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.8709294634702373e-07,
+      "loss": 1.6479,
+      "step": 1833
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.8601423792191523e-07,
+      "loss": 1.6356,
+      "step": 1834
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.849366750158305e-07,
+      "loss": 1.7057,
+      "step": 1835
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.8386025963918947e-07,
+      "loss": 1.6464,
+      "step": 1836
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.8278499380027006e-07,
+      "loss": 1.6445,
+      "step": 1837
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.817108795052061e-07,
+      "loss": 1.641,
+      "step": 1838
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.8063791875798243e-07,
+      "loss": 1.6598,
+      "step": 1839
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7956611356043187e-07,
+      "loss": 1.6582,
+      "step": 1840
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7849546591223227e-07,
+      "loss": 1.6585,
+      "step": 1841
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7742597781090056e-07,
+      "loss": 1.6498,
+      "step": 1842
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.763576512517904e-07,
+      "loss": 1.6605,
+      "step": 1843
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.752904882280898e-07,
+      "loss": 1.6793,
+      "step": 1844
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.742244907308135e-07,
+      "loss": 1.6616,
+      "step": 1845
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.731596607488041e-07,
+      "loss": 1.6411,
+      "step": 1846
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.720960002687247e-07,
+      "loss": 1.6786,
+      "step": 1847
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.710335112750561e-07,
+      "loss": 1.6995,
+      "step": 1848
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.699721957500953e-07,
+      "loss": 1.6787,
+      "step": 1849
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.6891205567394746e-07,
+      "loss": 1.675,
+      "step": 1850
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.6785309302452605e-07,
+      "loss": 1.6382,
+      "step": 1851
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.6679530977754835e-07,
+      "loss": 1.6481,
+      "step": 1852
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.6573870790653014e-07,
+      "loss": 1.6778,
+      "step": 1853
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.6468328938278316e-07,
+      "loss": 1.624,
+      "step": 1854
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.6362905617541274e-07,
+      "loss": 1.6696,
+      "step": 1855
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.625760102513102e-07,
+      "loss": 1.6335,
+      "step": 1856
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.6152415357515454e-07,
+      "loss": 1.6509,
+      "step": 1857
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.6047348810940426e-07,
+      "loss": 1.6463,
+      "step": 1858
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.5942401581429526e-07,
+      "loss": 1.6631,
+      "step": 1859
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.5837573864783886e-07,
+      "loss": 1.6551,
+      "step": 1860
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.573286585658154e-07,
+      "loss": 1.6719,
+      "step": 1861
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.5628277752177126e-07,
+      "loss": 1.6703,
+      "step": 1862
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.552380974670175e-07,
+      "loss": 1.6797,
+      "step": 1863
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.541946203506231e-07,
+      "loss": 1.6418,
+      "step": 1864
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.5315234811941296e-07,
+      "loss": 1.641,
+      "step": 1865
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.521112827179652e-07,
+      "loss": 1.65,
+      "step": 1866
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.5107142608860395e-07,
+      "loss": 1.6442,
+      "step": 1867
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.500327801714006e-07,
+      "loss": 1.6597,
+      "step": 1868
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.489953469041662e-07,
+      "loss": 1.6657,
+      "step": 1869
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.4795912822244955e-07,
+      "loss": 1.6808,
+      "step": 1870
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.4692412605953416e-07,
+      "loss": 1.657,
+      "step": 1871
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.458903423464331e-07,
+      "loss": 1.6724,
+      "step": 1872
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.448577790118863e-07,
+      "loss": 1.6822,
+      "step": 1873
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.43826437982357e-07,
+      "loss": 1.6615,
+      "step": 1874
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.427963211820274e-07,
+      "loss": 1.6769,
+      "step": 1875
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.41767430532797e-07,
+      "loss": 1.7027,
+      "step": 1876
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.407397679542763e-07,
+      "loss": 1.6556,
+      "step": 1877
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.3971333536378497e-07,
+      "loss": 1.6742,
+      "step": 1878
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.3868813467634827e-07,
+      "loss": 1.6562,
+      "step": 1879
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.376641678046925e-07,
+      "loss": 1.674,
+      "step": 1880
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.3664143665924214e-07,
+      "loss": 1.683,
+      "step": 1881
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.3561994314811693e-07,
+      "loss": 1.6275,
+      "step": 1882
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.345996891771267e-07,
+      "loss": 1.6665,
+      "step": 1883
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.335806766497686e-07,
+      "loss": 1.6884,
+      "step": 1884
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.325629074672244e-07,
+      "loss": 1.6647,
+      "step": 1885
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.3154638352835484e-07,
+      "loss": 1.6249,
+      "step": 1886
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.305311067296992e-07,
+      "loss": 1.6659,
+      "step": 1887
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.295170789654685e-07,
+      "loss": 1.6816,
+      "step": 1888
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.28504302127544e-07,
+      "loss": 1.6725,
+      "step": 1889
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.274927781054728e-07,
+      "loss": 1.6838,
+      "step": 1890
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.264825087864651e-07,
+      "loss": 1.6604,
+      "step": 1891
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.2547349605538933e-07,
+      "loss": 1.6464,
+      "step": 1892
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.244657417947707e-07,
+      "loss": 1.6467,
+      "step": 1893
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.234592478847856e-07,
+      "loss": 1.6709,
+      "step": 1894
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.224540162032593e-07,
+      "loss": 1.6327,
+      "step": 1895
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.214500486256618e-07,
+      "loss": 1.667,
+      "step": 1896
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.204473470251047e-07,
+      "loss": 1.6664,
+      "step": 1897
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.1944591327233837e-07,
+      "loss": 1.6654,
+      "step": 1898
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.184457492357471e-07,
+      "loss": 1.6538,
+      "step": 1899
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.174468567813461e-07,
+      "loss": 1.6867,
+      "step": 1900
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.1644923777277853e-07,
+      "loss": 1.6519,
+      "step": 1901
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.154528940713113e-07,
+      "loss": 1.6344,
+      "step": 1902
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.144578275358328e-07,
+      "loss": 1.6498,
+      "step": 1903
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.1346404002284786e-07,
+      "loss": 1.6322,
+      "step": 1904
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.124715333864748e-07,
+      "loss": 1.6751,
+      "step": 1905
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.114803094784435e-07,
+      "loss": 1.6708,
+      "step": 1906
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.10490370148089e-07,
+      "loss": 1.6673,
+      "step": 1907
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.095017172423502e-07,
+      "loss": 1.6819,
+      "step": 1908
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.0851435260576695e-07,
+      "loss": 1.6363,
+      "step": 1909
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.0752827808047443e-07,
+      "loss": 1.6931,
+      "step": 1910
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.0654349550620104e-07,
+      "loss": 1.6445,
+      "step": 1911
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.0556000672026516e-07,
+      "loss": 1.6448,
+      "step": 1912
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.045778135575705e-07,
+      "loss": 1.6565,
+      "step": 1913
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.035969178506048e-07,
+      "loss": 1.6685,
+      "step": 1914
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.026173214294343e-07,
+      "loss": 1.6722,
+      "step": 1915
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.016390261217008e-07,
+      "loss": 1.6665,
+      "step": 1916
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.0066203375262e-07,
+      "loss": 1.6724,
+      "step": 1917
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.996863461449749e-07,
+      "loss": 1.6811,
+      "step": 1918
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.98711965119115e-07,
+      "loss": 1.6586,
+      "step": 1919
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.977388924929529e-07,
+      "loss": 1.6489,
+      "step": 1920
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.9676713008195896e-07,
+      "loss": 1.6906,
+      "step": 1921
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.957966796991592e-07,
+      "loss": 1.6787,
+      "step": 1922
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.94827543155133e-07,
+      "loss": 1.6413,
+      "step": 1923
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.9385972225800625e-07,
+      "loss": 1.6141,
+      "step": 1924
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.9289321881345254e-07,
+      "loss": 1.6772,
+      "step": 1925
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.9192803462468607e-07,
+      "loss": 1.666,
+      "step": 1926
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.9096417149245966e-07,
+      "loss": 1.6475,
+      "step": 1927
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.9000163121506306e-07,
+      "loss": 1.6501,
+      "step": 1928
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.890404155883153e-07,
+      "loss": 1.6477,
+      "step": 1929
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.8808052640556634e-07,
+      "loss": 1.6928,
+      "step": 1930
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.871219654576903e-07,
+      "loss": 1.6677,
+      "step": 1931
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.86164734533083e-07,
+      "loss": 1.6549,
+      "step": 1932
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.852088354176597e-07,
+      "loss": 1.6593,
+      "step": 1933
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.842542698948507e-07,
+      "loss": 1.6661,
+      "step": 1934
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.833010397455966e-07,
+      "loss": 1.6806,
+      "step": 1935
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.8234914674834907e-07,
+      "loss": 1.6507,
+      "step": 1936
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.813985926790634e-07,
+      "loss": 1.6255,
+      "step": 1937
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.8044937931119707e-07,
+      "loss": 1.6598,
+      "step": 1938
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7950150841570683e-07,
+      "loss": 1.618,
+      "step": 1939
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.785549817610443e-07,
+      "loss": 1.6794,
+      "step": 1940
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7760980111315304e-07,
+      "loss": 1.6633,
+      "step": 1941
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.766659682354655e-07,
+      "loss": 1.6695,
+      "step": 1942
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.757234848888993e-07,
+      "loss": 1.6949,
+      "step": 1943
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.74782352831855e-07,
+      "loss": 1.6642,
+      "step": 1944
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.738425738202118e-07,
+      "loss": 1.6415,
+      "step": 1945
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7290414960732313e-07,
+      "loss": 1.6419,
+      "step": 1946
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7196708194401686e-07,
+      "loss": 1.6538,
+      "step": 1947
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7103137257858863e-07,
+      "loss": 1.6649,
+      "step": 1948
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.700970232568001e-07,
+      "loss": 1.6414,
+      "step": 1949
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6916403572187587e-07,
+      "loss": 1.6647,
+      "step": 1950
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6823241171449963e-07,
+      "loss": 1.6538,
+      "step": 1951
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6730215297281076e-07,
+      "loss": 1.6309,
+      "step": 1952
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.663732612324021e-07,
+      "loss": 1.6517,
+      "step": 1953
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.654457382263152e-07,
+      "loss": 1.6633,
+      "step": 1954
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.645195856850391e-07,
+      "loss": 1.6611,
+      "step": 1955
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.63594805336505e-07,
+      "loss": 1.6531,
+      "step": 1956
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.626713989060845e-07,
+      "loss": 1.6799,
+      "step": 1957
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.617493681165854e-07,
+      "loss": 1.6573,
+      "step": 1958
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.608287146882492e-07,
+      "loss": 1.6694,
+      "step": 1959
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.5990944033874805e-07,
+      "loss": 1.6546,
+      "step": 1960
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.589915467831807e-07,
+      "loss": 1.6812,
+      "step": 1961
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.5807503573406965e-07,
+      "loss": 1.6528,
+      "step": 1962
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.571599089013582e-07,
+      "loss": 1.653,
+      "step": 1963
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.5624616799240715e-07,
+      "loss": 1.6316,
+      "step": 1964
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.5533381471199135e-07,
+      "loss": 1.6292,
+      "step": 1965
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.544228507622972e-07,
+      "loss": 1.6306,
+      "step": 1966
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.535132778429188e-07,
+      "loss": 1.6466,
+      "step": 1967
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.526050976508547e-07,
+      "loss": 1.6634,
+      "step": 1968
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.516983118805053e-07,
+      "loss": 1.6309,
+      "step": 1969
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.50792922223669e-07,
+      "loss": 1.6789,
+      "step": 1970
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.498889303695404e-07,
+      "loss": 1.6627,
+      "step": 1971
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.489863380047055e-07,
+      "loss": 1.6724,
+      "step": 1972
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.480851468131393e-07,
+      "loss": 1.6629,
+      "step": 1973
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.471853584762026e-07,
+      "loss": 1.6795,
+      "step": 1974
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.4628697467263913e-07,
+      "loss": 1.6726,
+      "step": 1975
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.4538999707857156e-07,
+      "loss": 1.6664,
+      "step": 1976
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.4449442736750023e-07,
+      "loss": 1.6735,
+      "step": 1977
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.436002672102975e-07,
+      "loss": 1.6722,
+      "step": 1978
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.4270751827520654e-07,
+      "loss": 1.6585,
+      "step": 1979
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.418161822278374e-07,
+      "loss": 1.6873,
+      "step": 1980
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.4092626073116394e-07,
+      "loss": 1.665,
+      "step": 1981
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.400377554455215e-07,
+      "loss": 1.6691,
+      "step": 1982
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.391506680286026e-07,
+      "loss": 1.6603,
+      "step": 1983
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.3826500013545425e-07,
+      "loss": 1.6944,
+      "step": 1984
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.373807534184763e-07,
+      "loss": 1.682,
+      "step": 1985
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.364979295274151e-07,
+      "loss": 1.6605,
+      "step": 1986
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.356165301093641e-07,
+      "loss": 1.6765,
+      "step": 1987
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.347365568087586e-07,
+      "loss": 1.6562,
+      "step": 1988
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.3385801126737248e-07,
+      "loss": 1.6551,
+      "step": 1989
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.329808951243174e-07,
+      "loss": 1.6326,
+      "step": 1990
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.3210521001603634e-07,
+      "loss": 1.6483,
+      "step": 1991
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.312309575763034e-07,
+      "loss": 1.6527,
+      "step": 1992
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.3035813943621996e-07,
+      "loss": 1.6602,
+      "step": 1993
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.2948675722421085e-07,
+      "loss": 1.6562,
+      "step": 1994
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.2861681256602182e-07,
+      "loss": 1.6788,
+      "step": 1995
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.277483070847177e-07,
+      "loss": 1.6555,
+      "step": 1996
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.2688124240067619e-07,
+      "loss": 1.6549,
+      "step": 1997
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.2601562013158893e-07,
+      "loss": 1.6623,
+      "step": 1998
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.2515144189245505e-07,
+      "loss": 1.6464,
+      "step": 1999
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.2428870929558007e-07,
+      "loss": 1.6542,
+      "step": 2000
+    },
+    {
+      "epoch": 1.41,
+      "eval_loss": 1.6625683307647705,
+      "eval_runtime": 78.9188,
+      "eval_samples_per_second": 25.343,
+      "eval_steps_per_second": 0.063,
+      "step": 2000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.2342742395057267e-07,
+      "loss": 1.6468,
+      "step": 2001
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.2256758746434113e-07,
+      "loss": 1.638,
+      "step": 2002
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.2170920144108963e-07,
+      "loss": 1.6593,
+      "step": 2003
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.2085226748231788e-07,
+      "loss": 1.6636,
+      "step": 2004
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.1999678718681537e-07,
+      "loss": 1.6808,
+      "step": 2005
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.1914276215065953e-07,
+      "loss": 1.6514,
+      "step": 2006
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.182901939672137e-07,
+      "loss": 1.6512,
+      "step": 2007
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.1743908422712133e-07,
+      "loss": 1.656,
+      "step": 2008
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.1658943451830657e-07,
+      "loss": 1.6545,
+      "step": 2009
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.157412464259688e-07,
+      "loss": 1.6581,
+      "step": 2010
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.1489452153258015e-07,
+      "loss": 1.6641,
+      "step": 2011
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.1404926141788382e-07,
+      "loss": 1.6655,
+      "step": 2012
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.1320546765888926e-07,
+      "loss": 1.6441,
+      "step": 2013
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.1236314182987046e-07,
+      "loss": 1.668,
+      "step": 2014
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.1152228550236262e-07,
+      "loss": 1.6361,
+      "step": 2015
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.1068290024515922e-07,
+      "loss": 1.6629,
+      "step": 2016
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.0984498762430957e-07,
+      "loss": 1.6701,
+      "step": 2017
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.0900854920311505e-07,
+      "loss": 1.6554,
+      "step": 2018
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.0817358654212658e-07,
+      "loss": 1.6708,
+      "step": 2019
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.0734010119914192e-07,
+      "loss": 1.6589,
+      "step": 2020
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.065080947292025e-07,
+      "loss": 1.6673,
+      "step": 2021
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.0567756868459042e-07,
+      "loss": 1.6604,
+      "step": 2022
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.0484852461482639e-07,
+      "loss": 1.6655,
+      "step": 2023
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.0402096406666557e-07,
+      "loss": 1.6631,
+      "step": 2024
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.031948885840955e-07,
+      "loss": 1.6887,
+      "step": 2025
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.0237029970833287e-07,
+      "loss": 1.662,
+      "step": 2026
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.015471989778208e-07,
+      "loss": 1.6715,
+      "step": 2027
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.007255879282266e-07,
+      "loss": 1.6339,
+      "step": 2028
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9990546809243725e-07,
+      "loss": 1.6607,
+      "step": 2029
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.990868410005584e-07,
+      "loss": 1.6684,
+      "step": 2030
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9826970817991008e-07,
+      "loss": 1.6391,
+      "step": 2031
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9745407115502477e-07,
+      "loss": 1.6749,
+      "step": 2032
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.966399314476439e-07,
+      "loss": 1.6637,
+      "step": 2033
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9582729057671622e-07,
+      "loss": 1.6686,
+      "step": 2034
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9501615005839301e-07,
+      "loss": 1.6513,
+      "step": 2035
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9420651140602695e-07,
+      "loss": 1.6596,
+      "step": 2036
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.933983761301684e-07,
+      "loss": 1.6532,
+      "step": 2037
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9259174573856274e-07,
+      "loss": 1.6471,
+      "step": 2038
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9178662173614847e-07,
+      "loss": 1.6694,
+      "step": 2039
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9098300562505264e-07,
+      "loss": 1.6308,
+      "step": 2040
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9018089890458945e-07,
+      "loss": 1.6588,
+      "step": 2041
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.8938030307125708e-07,
+      "loss": 1.6657,
+      "step": 2042
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.8858121961873442e-07,
+      "loss": 1.6691,
+      "step": 2043
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.8778365003787932e-07,
+      "loss": 1.6653,
+      "step": 2044
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.8698759581672486e-07,
+      "loss": 1.6377,
+      "step": 2045
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.8619305844047662e-07,
+      "loss": 1.6553,
+      "step": 2046
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.8540003939151106e-07,
+      "loss": 1.6649,
+      "step": 2047
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.8460854014937067e-07,
+      "loss": 1.667,
+      "step": 2048
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.8381856219076308e-07,
+      "loss": 1.6854,
+      "step": 2049
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.83030106989558e-07,
+      "loss": 1.6716,
+      "step": 2050
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.8224317601678362e-07,
+      "loss": 1.6602,
+      "step": 2051
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.814577707406243e-07,
+      "loss": 1.6706,
+      "step": 2052
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.8067389262641819e-07,
+      "loss": 1.636,
+      "step": 2053
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7989154313665377e-07,
+      "loss": 1.6584,
+      "step": 2054
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7911072373096847e-07,
+      "loss": 1.661,
+      "step": 2055
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7833143586614408e-07,
+      "loss": 1.6532,
+      "step": 2056
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.77553680996105e-07,
+      "loss": 1.6573,
+      "step": 2057
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7677746057191667e-07,
+      "loss": 1.6811,
+      "step": 2058
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7600277604178003e-07,
+      "loss": 1.6324,
+      "step": 2059
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7522962885103143e-07,
+      "loss": 1.6289,
+      "step": 2060
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7445802044213932e-07,
+      "loss": 1.6385,
+      "step": 2061
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7368795225470045e-07,
+      "loss": 1.6643,
+      "step": 2062
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7291942572543805e-07,
+      "loss": 1.6829,
+      "step": 2063
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7215244228820024e-07,
+      "loss": 1.682,
+      "step": 2064
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7138700337395407e-07,
+      "loss": 1.6481,
+      "step": 2065
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7062311041078693e-07,
+      "loss": 1.6501,
+      "step": 2066
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6986076482390077e-07,
+      "loss": 1.652,
+      "step": 2067
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.69099968035611e-07,
+      "loss": 1.6388,
+      "step": 2068
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6834072146534384e-07,
+      "loss": 1.6543,
+      "step": 2069
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6758302652963174e-07,
+      "loss": 1.6695,
+      "step": 2070
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6682688464211425e-07,
+      "loss": 1.6669,
+      "step": 2071
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.66072297213532e-07,
+      "loss": 1.6444,
+      "step": 2072
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.653192656517257e-07,
+      "loss": 1.6584,
+      "step": 2073
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6456779136163402e-07,
+      "loss": 1.6601,
+      "step": 2074
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6381787574528938e-07,
+      "loss": 1.6811,
+      "step": 2075
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6306952020181573e-07,
+      "loss": 1.6655,
+      "step": 2076
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6232272612742804e-07,
+      "loss": 1.6734,
+      "step": 2077
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.615774949154266e-07,
+      "loss": 1.6412,
+      "step": 2078
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6083382795619604e-07,
+      "loss": 1.6572,
+      "step": 2079
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.600917266372035e-07,
+      "loss": 1.6685,
+      "step": 2080
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5935119234299389e-07,
+      "loss": 1.6875,
+      "step": 2081
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5861222645518922e-07,
+      "loss": 1.6606,
+      "step": 2082
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5787483035248516e-07,
+      "loss": 1.6367,
+      "step": 2083
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5713900541064817e-07,
+      "loss": 1.6846,
+      "step": 2084
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.564047530025142e-07,
+      "loss": 1.6614,
+      "step": 2085
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5567207449798515e-07,
+      "loss": 1.6477,
+      "step": 2086
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.549409712640253e-07,
+      "loss": 1.6641,
+      "step": 2087
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5421144466466162e-07,
+      "loss": 1.6454,
+      "step": 2088
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5348349606097854e-07,
+      "loss": 1.6644,
+      "step": 2089
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.527571268111164e-07,
+      "loss": 1.6602,
+      "step": 2090
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5203233827026963e-07,
+      "loss": 1.6504,
+      "step": 2091
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5130913179068272e-07,
+      "loss": 1.6442,
+      "step": 2092
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5058750872164882e-07,
+      "loss": 1.654,
+      "step": 2093
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.498674704095071e-07,
+      "loss": 1.6537,
+      "step": 2094
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4914901819763936e-07,
+      "loss": 1.6937,
+      "step": 2095
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.484321534264692e-07,
+      "loss": 1.6571,
+      "step": 2096
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4771687743345784e-07,
+      "loss": 1.6733,
+      "step": 2097
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4700319155310226e-07,
+      "loss": 1.6531,
+      "step": 2098
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4629109711693311e-07,
+      "loss": 1.7078,
+      "step": 2099
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4558059545351142e-07,
+      "loss": 1.6526,
+      "step": 2100
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.448716878884272e-07,
+      "loss": 1.6425,
+      "step": 2101
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4416437574429585e-07,
+      "loss": 1.6595,
+      "step": 2102
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4345866034075627e-07,
+      "loss": 1.6563,
+      "step": 2103
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.427545429944683e-07,
+      "loss": 1.6603,
+      "step": 2104
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.420520250191105e-07,
+      "loss": 1.6459,
+      "step": 2105
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4135110772537683e-07,
+      "loss": 1.6636,
+      "step": 2106
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.406517924209758e-07,
+      "loss": 1.6086,
+      "step": 2107
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.399540804106264e-07,
+      "loss": 1.6644,
+      "step": 2108
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.3925797299605645e-07,
+      "loss": 1.6577,
+      "step": 2109
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.3856347147600013e-07,
+      "loss": 1.6684,
+      "step": 2110
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.3787057714619532e-07,
+      "loss": 1.6721,
+      "step": 2111
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.3717929129938177e-07,
+      "loss": 1.6444,
+      "step": 2112
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.36489615225298e-07,
+      "loss": 1.6477,
+      "step": 2113
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.3580155021067895e-07,
+      "loss": 1.662,
+      "step": 2114
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.3511509753925422e-07,
+      "loss": 1.6431,
+      "step": 2115
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.3443025849174482e-07,
+      "loss": 1.6656,
+      "step": 2116
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.3374703434586132e-07,
+      "loss": 1.6597,
+      "step": 2117
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.3306542637630192e-07,
+      "loss": 1.6741,
+      "step": 2118
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.323854358547487e-07,
+      "loss": 1.6472,
+      "step": 2119
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.3170706404986641e-07,
+      "loss": 1.6599,
+      "step": 2120
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.3103031222729978e-07,
+      "loss": 1.6782,
+      "step": 2121
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.3035518164967086e-07,
+      "loss": 1.6591,
+      "step": 2122
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.2968167357657743e-07,
+      "loss": 1.6692,
+      "step": 2123
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.2900978926458984e-07,
+      "loss": 1.6603,
+      "step": 2124
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.283395299672486e-07,
+      "loss": 1.6491,
+      "step": 2125
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.276708969350636e-07,
+      "loss": 1.6484,
+      "step": 2126
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.2700389141550883e-07,
+      "loss": 1.6551,
+      "step": 2127
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2633851465302337e-07,
+      "loss": 1.6485,
+      "step": 2128
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2567476788900676e-07,
+      "loss": 1.635,
+      "step": 2129
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2501265236181734e-07,
+      "loss": 1.6682,
+      "step": 2130
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2435216930677107e-07,
+      "loss": 1.6449,
+      "step": 2131
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2369331995613663e-07,
+      "loss": 1.6389,
+      "step": 2132
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2303610553913547e-07,
+      "loss": 1.6654,
+      "step": 2133
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2238052728193925e-07,
+      "loss": 1.6969,
+      "step": 2134
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.217265864076662e-07,
+      "loss": 1.6803,
+      "step": 2135
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2107428413637976e-07,
+      "loss": 1.6849,
+      "step": 2136
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2042362168508713e-07,
+      "loss": 1.6858,
+      "step": 2137
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.1977460026773444e-07,
+      "loss": 1.6496,
+      "step": 2138
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.1912722109520789e-07,
+      "loss": 1.6702,
+      "step": 2139
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.1848148537532842e-07,
+      "loss": 1.6611,
+      "step": 2140
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.1783739431285122e-07,
+      "loss": 1.6389,
+      "step": 2141
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.1719494910946348e-07,
+      "loss": 1.6596,
+      "step": 2142
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.1655415096378118e-07,
+      "loss": 1.652,
+      "step": 2143
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.15915001071347e-07,
+      "loss": 1.6697,
+      "step": 2144
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.1527750062462926e-07,
+      "loss": 1.6595,
+      "step": 2145
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.1464165081301858e-07,
+      "loss": 1.6661,
+      "step": 2146
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.1400745282282564e-07,
+      "loss": 1.6472,
+      "step": 2147
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.1337490783728032e-07,
+      "loss": 1.6544,
+      "step": 2148
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.1274401703652692e-07,
+      "loss": 1.6476,
+      "step": 2149
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.1211478159762478e-07,
+      "loss": 1.6646,
+      "step": 2150
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.1148720269454448e-07,
+      "loss": 1.649,
+      "step": 2151
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.1086128149816542e-07,
+      "loss": 1.6705,
+      "step": 2152
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.1023701917627526e-07,
+      "loss": 1.6536,
+      "step": 2153
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0961441689356587e-07,
+      "loss": 1.6838,
+      "step": 2154
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.089934758116322e-07,
+      "loss": 1.6689,
+      "step": 2155
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0837419708896977e-07,
+      "loss": 1.6217,
+      "step": 2156
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.077565818809727e-07,
+      "loss": 1.6531,
+      "step": 2157
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0714063133993178e-07,
+      "loss": 1.6543,
+      "step": 2158
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0652634661503168e-07,
+      "loss": 1.642,
+      "step": 2159
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0591372885234884e-07,
+      "loss": 1.6412,
+      "step": 2160
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0530277919485042e-07,
+      "loss": 1.6252,
+      "step": 2161
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0469349878239076e-07,
+      "loss": 1.6694,
+      "step": 2162
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0408588875170987e-07,
+      "loss": 1.6803,
+      "step": 2163
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0347995023643196e-07,
+      "loss": 1.6392,
+      "step": 2164
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0287568436706206e-07,
+      "loss": 1.7027,
+      "step": 2165
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0227309227098468e-07,
+      "loss": 1.6775,
+      "step": 2166
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0167217507246151e-07,
+      "loss": 1.6696,
+      "step": 2167
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0107293389262917e-07,
+      "loss": 1.6641,
+      "step": 2168
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0047536984949811e-07,
+      "loss": 1.6486,
+      "step": 2169
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.987948405794911e-08,
+      "loss": 1.6578,
+      "step": 2170
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.92852776297316e-08,
+      "loss": 1.6367,
+      "step": 2171
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.869275167346236e-08,
+      "loss": 1.5966,
+      "step": 2172
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.810190729462254e-08,
+      "loss": 1.6714,
+      "step": 2173
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.751274559555589e-08,
+      "loss": 1.6631,
+      "step": 2174
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.692526767546728e-08,
+      "loss": 1.6677,
+      "step": 2175
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.633947463041981e-08,
+      "loss": 1.6728,
+      "step": 2176
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.57553675533328e-08,
+      "loss": 1.6689,
+      "step": 2177
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.517294753398064e-08,
+      "loss": 1.6717,
+      "step": 2178
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.459221565898933e-08,
+      "loss": 1.6746,
+      "step": 2179
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.401317301183653e-08,
+      "loss": 1.6767,
+      "step": 2180
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.343582067284705e-08,
+      "loss": 1.6454,
+      "step": 2181
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.286015971919281e-08,
+      "loss": 1.6282,
+      "step": 2182
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.228619122488979e-08,
+      "loss": 1.6551,
+      "step": 2183
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.171391626079628e-08,
+      "loss": 1.6636,
+      "step": 2184
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.114333589461143e-08,
+      "loss": 1.6539,
+      "step": 2185
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.057445119087215e-08,
+      "loss": 1.6531,
+      "step": 2186
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.00072632109521e-08,
+      "loss": 1.6874,
+      "step": 2187
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.94417730130591e-08,
+      "loss": 1.6443,
+      "step": 2188
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.887798165223381e-08,
+      "loss": 1.6542,
+      "step": 2189
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.831589018034657e-08,
+      "loss": 1.6908,
+      "step": 2190
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.775549964609741e-08,
+      "loss": 1.6582,
+      "step": 2191
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.719681109501175e-08,
+      "loss": 1.665,
+      "step": 2192
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.663982556944039e-08,
+      "loss": 1.6835,
+      "step": 2193
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.608454410855625e-08,
+      "loss": 1.6433,
+      "step": 2194
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.55309677483531e-08,
+      "loss": 1.6463,
+      "step": 2195
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.49790975216439e-08,
+      "loss": 1.6851,
+      "step": 2196
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.4428934458058e-08,
+      "loss": 1.6943,
+      "step": 2197
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.388047958403954e-08,
+      "loss": 1.6416,
+      "step": 2198
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.33337339228466e-08,
+      "loss": 1.612,
+      "step": 2199
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.278869849454717e-08,
+      "loss": 1.6688,
+      "step": 2200
+    },
+    {
+      "epoch": 1.55,
+      "eval_loss": 1.6619179248809814,
+      "eval_runtime": 75.1957,
+      "eval_samples_per_second": 26.597,
+      "eval_steps_per_second": 0.066,
+      "step": 2200
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.16596771652649e-07,
+      "loss": 1.6952,
+      "step": 2201
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.157729457829502e-07,
+      "loss": 1.6372,
+      "step": 2202
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.149491774816491e-07,
+      "loss": 1.6743,
+      "step": 2203
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.141254673117829e-07,
+      "loss": 1.6598,
+      "step": 2204
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.133018158363502e-07,
+      "loss": 1.665,
+      "step": 2205
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.124782236183087e-07,
+      "loss": 1.6883,
+      "step": 2206
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.116546912205763e-07,
+      "loss": 1.6477,
+      "step": 2207
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.108312192060296e-07,
+      "loss": 1.6525,
+      "step": 2208
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.100078081375034e-07,
+      "loss": 1.6499,
+      "step": 2209
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.091844585777917e-07,
+      "loss": 1.6394,
+      "step": 2210
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.083611710896465e-07,
+      "loss": 1.6365,
+      "step": 2211
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.075379462357765e-07,
+      "loss": 1.6541,
+      "step": 2212
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.067147845788485e-07,
+      "loss": 1.6679,
+      "step": 2213
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.058916866814856e-07,
+      "loss": 1.6384,
+      "step": 2214
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.050686531062676e-07,
+      "loss": 1.6851,
+      "step": 2215
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.042456844157298e-07,
+      "loss": 1.6612,
+      "step": 2216
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.034227811723642e-07,
+      "loss": 1.6802,
+      "step": 2217
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.025999439386169e-07,
+      "loss": 1.6499,
+      "step": 2218
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.01777173276889e-07,
+      "loss": 1.6485,
+      "step": 2219
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.009544697495373e-07,
+      "loss": 1.6395,
+      "step": 2220
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.001318339188716e-07,
+      "loss": 1.6443,
+      "step": 2221
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.993092663471556e-07,
+      "loss": 1.6529,
+      "step": 2222
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.984867675966068e-07,
+      "loss": 1.6487,
+      "step": 2223
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.97664338229395e-07,
+      "loss": 1.6568,
+      "step": 2224
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.96841978807643e-07,
+      "loss": 1.6427,
+      "step": 2225
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.960196898934262e-07,
+      "loss": 1.6854,
+      "step": 2226
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.951974720487704e-07,
+      "loss": 1.6589,
+      "step": 2227
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.943753258356545e-07,
+      "loss": 1.6646,
+      "step": 2228
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.93553251816007e-07,
+      "loss": 1.6378,
+      "step": 2229
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.927312505517084e-07,
+      "loss": 1.6756,
+      "step": 2230
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.919093226045886e-07,
+      "loss": 1.6591,
+      "step": 2231
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.910874685364273e-07,
+      "loss": 1.6856,
+      "step": 2232
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.902656889089547e-07,
+      "loss": 1.6653,
+      "step": 2233
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.894439842838487e-07,
+      "loss": 1.664,
+      "step": 2234
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.886223552227372e-07,
+      "loss": 1.6515,
+      "step": 2235
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.878008022871958e-07,
+      "loss": 1.6724,
+      "step": 2236
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.869793260387477e-07,
+      "loss": 1.6743,
+      "step": 2237
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.861579270388647e-07,
+      "loss": 1.6504,
+      "step": 2238
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.85336605848965e-07,
+      "loss": 1.6641,
+      "step": 2239
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.845153630304139e-07,
+      "loss": 1.675,
+      "step": 2240
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.836941991445233e-07,
+      "loss": 1.6537,
+      "step": 2241
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.828731147525509e-07,
+      "loss": 1.6835,
+      "step": 2242
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.820521104157003e-07,
+      "loss": 1.6412,
+      "step": 2243
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.812311866951198e-07,
+      "loss": 1.6213,
+      "step": 2244
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.804103441519036e-07,
+      "loss": 1.655,
+      "step": 2245
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.795895833470895e-07,
+      "loss": 1.6528,
+      "step": 2246
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.787689048416593e-07,
+      "loss": 1.655,
+      "step": 2247
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.779483091965398e-07,
+      "loss": 1.6604,
+      "step": 2248
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.771277969725999e-07,
+      "loss": 1.6553,
+      "step": 2249
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.763073687306522e-07,
+      "loss": 1.6342,
+      "step": 2250
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.754870250314519e-07,
+      "loss": 1.6375,
+      "step": 2251
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.746667664356955e-07,
+      "loss": 1.6522,
+      "step": 2252
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.738465935040229e-07,
+      "loss": 1.6819,
+      "step": 2253
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.73026506797014e-07,
+      "loss": 1.6591,
+      "step": 2254
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.722065068751903e-07,
+      "loss": 1.6622,
+      "step": 2255
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.713865942990141e-07,
+      "loss": 1.6785,
+      "step": 2256
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.705667696288878e-07,
+      "loss": 1.646,
+      "step": 2257
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.697470334251542e-07,
+      "loss": 1.6701,
+      "step": 2258
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.689273862480948e-07,
+      "loss": 1.6363,
+      "step": 2259
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.68107828657931e-07,
+      "loss": 1.6673,
+      "step": 2260
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.672883612148229e-07,
+      "loss": 1.6508,
+      "step": 2261
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.664689844788682e-07,
+      "loss": 1.6587,
+      "step": 2262
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.656496990101038e-07,
+      "loss": 1.6904,
+      "step": 2263
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.648305053685034e-07,
+      "loss": 1.6733,
+      "step": 2264
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.640114041139776e-07,
+      "loss": 1.6735,
+      "step": 2265
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.631923958063749e-07,
+      "loss": 1.6599,
+      "step": 2266
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.623734810054796e-07,
+      "loss": 1.6535,
+      "step": 2267
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.615546602710124e-07,
+      "loss": 1.6495,
+      "step": 2268
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.607359341626294e-07,
+      "loss": 1.6497,
+      "step": 2269
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.599173032399221e-07,
+      "loss": 1.6495,
+      "step": 2270
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.590987680624174e-07,
+      "loss": 1.6637,
+      "step": 2271
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.582803291895757e-07,
+      "loss": 1.6468,
+      "step": 2272
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.574619871807934e-07,
+      "loss": 1.6673,
+      "step": 2273
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.566437425953982e-07,
+      "loss": 1.6695,
+      "step": 2274
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.558255959926533e-07,
+      "loss": 1.6801,
+      "step": 2275
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.550075479317542e-07,
+      "loss": 1.633,
+      "step": 2276
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.541895989718286e-07,
+      "loss": 1.6575,
+      "step": 2277
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.533717496719372e-07,
+      "loss": 1.6401,
+      "step": 2278
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.525540005910725e-07,
+      "loss": 1.6943,
+      "step": 2279
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.517363522881578e-07,
+      "loss": 1.6701,
+      "step": 2280
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.509188053220486e-07,
+      "loss": 1.6462,
+      "step": 2281
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.501013602515304e-07,
+      "loss": 1.6245,
+      "step": 2282
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.492840176353185e-07,
+      "loss": 1.6505,
+      "step": 2283
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.484667780320597e-07,
+      "loss": 1.65,
+      "step": 2284
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.47649642000329e-07,
+      "loss": 1.6704,
+      "step": 2285
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.468326100986316e-07,
+      "loss": 1.6678,
+      "step": 2286
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.460156828854007e-07,
+      "loss": 1.6684,
+      "step": 2287
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.451988609189986e-07,
+      "loss": 1.6558,
+      "step": 2288
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.443821447577155e-07,
+      "loss": 1.6442,
+      "step": 2289
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.435655349597689e-07,
+      "loss": 1.6598,
+      "step": 2290
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.427490320833046e-07,
+      "loss": 1.6375,
+      "step": 2291
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.419326366863937e-07,
+      "loss": 1.6375,
+      "step": 2292
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.41116349327035e-07,
+      "loss": 1.6206,
+      "step": 2293
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.403001705631536e-07,
+      "loss": 1.6723,
+      "step": 2294
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.394841009525995e-07,
+      "loss": 1.6204,
+      "step": 2295
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.38668141053149e-07,
+      "loss": 1.6558,
+      "step": 2296
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.378522914225027e-07,
+      "loss": 1.6496,
+      "step": 2297
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.370365526182861e-07,
+      "loss": 1.6556,
+      "step": 2298
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.362209251980495e-07,
+      "loss": 1.6328,
+      "step": 2299
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.354054097192659e-07,
+      "loss": 1.6612,
+      "step": 2300
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.345900067393332e-07,
+      "loss": 1.6797,
+      "step": 2301
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.337747168155709e-07,
+      "loss": 1.6599,
+      "step": 2302
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.32959540505222e-07,
+      "loss": 1.6793,
+      "step": 2303
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.321444783654523e-07,
+      "loss": 1.6637,
+      "step": 2304
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.313295309533487e-07,
+      "loss": 1.6478,
+      "step": 2305
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.305146988259204e-07,
+      "loss": 1.6624,
+      "step": 2306
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.296999825400971e-07,
+      "loss": 1.6538,
+      "step": 2307
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.288853826527299e-07,
+      "loss": 1.6599,
+      "step": 2308
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.280708997205903e-07,
+      "loss": 1.6669,
+      "step": 2309
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.272565343003699e-07,
+      "loss": 1.6341,
+      "step": 2310
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.264422869486789e-07,
+      "loss": 1.6481,
+      "step": 2311
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.256281582220485e-07,
+      "loss": 1.6569,
+      "step": 2312
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.248141486769273e-07,
+      "loss": 1.661,
+      "step": 2313
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.24000258869684e-07,
+      "loss": 1.6266,
+      "step": 2314
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.23186489356604e-07,
+      "loss": 1.6073,
+      "step": 2315
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.223728406938913e-07,
+      "loss": 1.6509,
+      "step": 2316
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.21559313437667e-07,
+      "loss": 1.6766,
+      "step": 2317
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.207459081439694e-07,
+      "loss": 1.652,
+      "step": 2318
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.199326253687536e-07,
+      "loss": 1.6509,
+      "step": 2319
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.191194656678904e-07,
+      "loss": 1.6575,
+      "step": 2320
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.183064295971665e-07,
+      "loss": 1.6651,
+      "step": 2321
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.174935177122848e-07,
+      "loss": 1.6681,
+      "step": 2322
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.166807305688628e-07,
+      "loss": 1.6349,
+      "step": 2323
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.158680687224328e-07,
+      "loss": 1.6692,
+      "step": 2324
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.150555327284417e-07,
+      "loss": 1.6623,
+      "step": 2325
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.1424312314225e-07,
+      "loss": 1.6388,
+      "step": 2326
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.13430840519132e-07,
+      "loss": 1.6689,
+      "step": 2327
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.126186854142751e-07,
+      "loss": 1.6805,
+      "step": 2328
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.118066583827803e-07,
+      "loss": 1.637,
+      "step": 2329
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.109947599796599e-07,
+      "loss": 1.6806,
+      "step": 2330
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.101829907598384e-07,
+      "loss": 1.6317,
+      "step": 2331
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.093713512781532e-07,
+      "loss": 1.6373,
+      "step": 2332
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.085598420893516e-07,
+      "loss": 1.7101,
+      "step": 2333
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.07748463748093e-07,
+      "loss": 1.6679,
+      "step": 2334
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.069372168089465e-07,
+      "loss": 1.6673,
+      "step": 2335
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.061261018263918e-07,
+      "loss": 1.6706,
+      "step": 2336
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.053151193548183e-07,
+      "loss": 1.6069,
+      "step": 2337
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.045042699485255e-07,
+      "loss": 1.6697,
+      "step": 2338
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.036935541617202e-07,
+      "loss": 1.6677,
+      "step": 2339
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.028829725485197e-07,
+      "loss": 1.637,
+      "step": 2340
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.020725256629488e-07,
+      "loss": 1.6569,
+      "step": 2341
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.012622140589404e-07,
+      "loss": 1.6577,
+      "step": 2342
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.004520382903344e-07,
+      "loss": 1.6652,
+      "step": 2343
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.996419989108788e-07,
+      "loss": 1.6468,
+      "step": 2344
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.988320964742275e-07,
+      "loss": 1.6593,
+      "step": 2345
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.980223315339417e-07,
+      "loss": 1.6265,
+      "step": 2346
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.972127046434877e-07,
+      "loss": 1.6521,
+      "step": 2347
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.964032163562377e-07,
+      "loss": 1.6657,
+      "step": 2348
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.955938672254695e-07,
+      "loss": 1.6464,
+      "step": 2349
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.947846578043658e-07,
+      "loss": 1.679,
+      "step": 2350
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.939755886460132e-07,
+      "loss": 1.6444,
+      "step": 2351
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.931666603034032e-07,
+      "loss": 1.6608,
+      "step": 2352
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.923578733294305e-07,
+      "loss": 1.6503,
+      "step": 2353
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.915492282768936e-07,
+      "loss": 1.6699,
+      "step": 2354
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.907407256984934e-07,
+      "loss": 1.6672,
+      "step": 2355
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.899323661468343e-07,
+      "loss": 1.6785,
+      "step": 2356
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.891241501744221e-07,
+      "loss": 1.6672,
+      "step": 2357
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.883160783336644e-07,
+      "loss": 1.6644,
+      "step": 2358
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.875081511768713e-07,
+      "loss": 1.6565,
+      "step": 2359
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.867003692562533e-07,
+      "loss": 1.6506,
+      "step": 2360
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.858927331239213e-07,
+      "loss": 1.652,
+      "step": 2361
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.850852433318877e-07,
+      "loss": 1.6498,
+      "step": 2362
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.842779004320634e-07,
+      "loss": 1.6157,
+      "step": 2363
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.834707049762603e-07,
+      "loss": 1.6743,
+      "step": 2364
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.826636575161884e-07,
+      "loss": 1.6671,
+      "step": 2365
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.818567586034576e-07,
+      "loss": 1.6532,
+      "step": 2366
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.810500087895752e-07,
+      "loss": 1.6541,
+      "step": 2367
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.802434086259468e-07,
+      "loss": 1.6316,
+      "step": 2368
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.794369586638767e-07,
+      "loss": 1.6606,
+      "step": 2369
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.786306594545656e-07,
+      "loss": 1.6697,
+      "step": 2370
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.778245115491112e-07,
+      "loss": 1.6714,
+      "step": 2371
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.770185154985084e-07,
+      "loss": 1.6688,
+      "step": 2372
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.762126718536475e-07,
+      "loss": 1.6545,
+      "step": 2373
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.754069811653154e-07,
+      "loss": 1.6456,
+      "step": 2374
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.746014439841941e-07,
+      "loss": 1.6664,
+      "step": 2375
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.737960608608599e-07,
+      "loss": 1.6552,
+      "step": 2376
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.729908323457855e-07,
+      "loss": 1.6715,
+      "step": 2377
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.721857589893363e-07,
+      "loss": 1.65,
+      "step": 2378
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.713808413417726e-07,
+      "loss": 1.6472,
+      "step": 2379
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.705760799532485e-07,
+      "loss": 1.6724,
+      "step": 2380
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.6977147537381e-07,
+      "loss": 1.6667,
+      "step": 2381
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.689670281533976e-07,
+      "loss": 1.6617,
+      "step": 2382
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.681627388418426e-07,
+      "loss": 1.6698,
+      "step": 2383
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.673586079888697e-07,
+      "loss": 1.6866,
+      "step": 2384
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.665546361440949e-07,
+      "loss": 1.6773,
+      "step": 2385
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.657508238570245e-07,
+      "loss": 1.6524,
+      "step": 2386
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.649471716770573e-07,
+      "loss": 1.6547,
+      "step": 2387
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.641436801534817e-07,
+      "loss": 1.6539,
+      "step": 2388
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.633403498354768e-07,
+      "loss": 1.6259,
+      "step": 2389
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.625371812721114e-07,
+      "loss": 1.6706,
+      "step": 2390
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.617341750123433e-07,
+      "loss": 1.6371,
+      "step": 2391
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.609313316050199e-07,
+      "loss": 1.6576,
+      "step": 2392
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.60128651598877e-07,
+      "loss": 1.632,
+      "step": 2393
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.593261355425393e-07,
+      "loss": 1.6525,
+      "step": 2394
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.585237839845183e-07,
+      "loss": 1.67,
+      "step": 2395
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.577215974732138e-07,
+      "loss": 1.6502,
+      "step": 2396
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.569195765569129e-07,
+      "loss": 1.6224,
+      "step": 2397
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.561177217837893e-07,
+      "loss": 1.6644,
+      "step": 2398
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.553160337019031e-07,
+      "loss": 1.6723,
+      "step": 2399
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.545145128592008e-07,
+      "loss": 1.6416,
+      "step": 2400
+    },
+    {
+      "epoch": 1.69,
+      "eval_loss": 1.6600674390792847,
+      "eval_runtime": 45.2976,
+      "eval_samples_per_second": 44.152,
+      "eval_steps_per_second": 0.11,
+      "step": 2400
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.53713159803514e-07,
+      "loss": 1.6299,
+      "step": 2401
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.529119750825605e-07,
+      "loss": 1.6389,
+      "step": 2402
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.521109592439422e-07,
+      "loss": 1.6675,
+      "step": 2403
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.513101128351453e-07,
+      "loss": 1.6391,
+      "step": 2404
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.505094364035416e-07,
+      "loss": 1.6429,
+      "step": 2405
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.497089304963853e-07,
+      "loss": 1.6744,
+      "step": 2406
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.48908595660815e-07,
+      "loss": 1.6734,
+      "step": 2407
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.48108432443852e-07,
+      "loss": 1.6509,
+      "step": 2408
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.473084413924e-07,
+      "loss": 1.6642,
+      "step": 2409
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.465086230532459e-07,
+      "loss": 1.6539,
+      "step": 2410
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.457089779730574e-07,
+      "loss": 1.6487,
+      "step": 2411
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.449095066983848e-07,
+      "loss": 1.6732,
+      "step": 2412
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.441102097756593e-07,
+      "loss": 1.6287,
+      "step": 2413
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.433110877511922e-07,
+      "loss": 1.6544,
+      "step": 2414
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.425121411711761e-07,
+      "loss": 1.6747,
+      "step": 2415
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.417133705816836e-07,
+      "loss": 1.6793,
+      "step": 2416
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.409147765286667e-07,
+      "loss": 1.6436,
+      "step": 2417
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.401163595579572e-07,
+      "loss": 1.6618,
+      "step": 2418
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.393181202152652e-07,
+      "loss": 1.6647,
+      "step": 2419
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.385200590461802e-07,
+      "loss": 1.6554,
+      "step": 2420
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.37722176596169e-07,
+      "loss": 1.6522,
+      "step": 2421
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.369244734105772e-07,
+      "loss": 1.6356,
+      "step": 2422
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.361269500346273e-07,
+      "loss": 1.6825,
+      "step": 2423
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.353296070134185e-07,
+      "loss": 1.6592,
+      "step": 2424
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.345324448919279e-07,
+      "loss": 1.6739,
+      "step": 2425
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.337354642150076e-07,
+      "loss": 1.6542,
+      "step": 2426
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.32938665527387e-07,
+      "loss": 1.6617,
+      "step": 2427
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.321420493736704e-07,
+      "loss": 1.6802,
+      "step": 2428
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.313456162983372e-07,
+      "loss": 1.6554,
+      "step": 2429
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.305493668457419e-07,
+      "loss": 1.6392,
+      "step": 2430
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.297533015601141e-07,
+      "loss": 1.6527,
+      "step": 2431
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.289574209855559e-07,
+      "loss": 1.6329,
+      "step": 2432
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.281617256660448e-07,
+      "loss": 1.6512,
+      "step": 2433
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.273662161454308e-07,
+      "loss": 1.6596,
+      "step": 2434
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.265708929674375e-07,
+      "loss": 1.662,
+      "step": 2435
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.257757566756603e-07,
+      "loss": 1.6563,
+      "step": 2436
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.249808078135677e-07,
+      "loss": 1.6751,
+      "step": 2437
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.241860469245e-07,
+      "loss": 1.6634,
+      "step": 2438
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.23391474551668e-07,
+      "loss": 1.6324,
+      "step": 2439
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.225970912381556e-07,
+      "loss": 1.6617,
+      "step": 2440
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.218028975269151e-07,
+      "loss": 1.6661,
+      "step": 2441
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.210088939607708e-07,
+      "loss": 1.6824,
+      "step": 2442
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.20215081082417e-07,
+      "loss": 1.6668,
+      "step": 2443
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.194214594344168e-07,
+      "loss": 1.6326,
+      "step": 2444
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.186280295592035e-07,
+      "loss": 1.642,
+      "step": 2445
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.178347919990788e-07,
+      "loss": 1.6493,
+      "step": 2446
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.170417472962129e-07,
+      "loss": 1.6856,
+      "step": 2447
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.162488959926449e-07,
+      "loss": 1.6447,
+      "step": 2448
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.154562386302808e-07,
+      "loss": 1.6443,
+      "step": 2449
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.146637757508949e-07,
+      "loss": 1.6465,
+      "step": 2450
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.138715078961272e-07,
+      "loss": 1.6715,
+      "step": 2451
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.130794356074858e-07,
+      "loss": 1.6686,
+      "step": 2452
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.12287559426345e-07,
+      "loss": 1.6429,
+      "step": 2453
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.114958798939441e-07,
+      "loss": 1.6377,
+      "step": 2454
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.10704397551389e-07,
+      "loss": 1.6295,
+      "step": 2455
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.099131129396501e-07,
+      "loss": 1.6595,
+      "step": 2456
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.091220265995631e-07,
+      "loss": 1.6481,
+      "step": 2457
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.083311390718284e-07,
+      "loss": 1.6682,
+      "step": 2458
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.0754045089701e-07,
+      "loss": 1.6883,
+      "step": 2459
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.067499626155354e-07,
+      "loss": 1.6815,
+      "step": 2460
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.059596747676962e-07,
+      "loss": 1.6791,
+      "step": 2461
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.051695878936466e-07,
+      "loss": 1.6335,
+      "step": 2462
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.043797025334037e-07,
+      "loss": 1.6448,
+      "step": 2463
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.035900192268464e-07,
+      "loss": 1.6508,
+      "step": 2464
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.028005385137161e-07,
+      "loss": 1.6294,
+      "step": 2465
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.020112609336153e-07,
+      "loss": 1.6375,
+      "step": 2466
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.012221870260077e-07,
+      "loss": 1.6542,
+      "step": 2467
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.004333173302184e-07,
+      "loss": 1.622,
+      "step": 2468
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.996446523854316e-07,
+      "loss": 1.6626,
+      "step": 2469
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.988561927306927e-07,
+      "loss": 1.6374,
+      "step": 2470
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.980679389049065e-07,
+      "loss": 1.6607,
+      "step": 2471
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.972798914468369e-07,
+      "loss": 1.653,
+      "step": 2472
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.964920508951071e-07,
+      "loss": 1.6549,
+      "step": 2473
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.957044177881986e-07,
+      "loss": 1.6772,
+      "step": 2474
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.949169926644512e-07,
+      "loss": 1.6289,
+      "step": 2475
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.941297760620626e-07,
+      "loss": 1.6515,
+      "step": 2476
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.933427685190875e-07,
+      "loss": 1.6531,
+      "step": 2477
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.925559705734393e-07,
+      "loss": 1.6526,
+      "step": 2478
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.917693827628859e-07,
+      "loss": 1.6655,
+      "step": 2479
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.909830056250526e-07,
+      "loss": 1.6566,
+      "step": 2480
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.901968396974214e-07,
+      "loss": 1.6606,
+      "step": 2481
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.894108855173289e-07,
+      "loss": 1.6414,
+      "step": 2482
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.886251436219674e-07,
+      "loss": 1.6589,
+      "step": 2483
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.87839614548384e-07,
+      "loss": 1.6354,
+      "step": 2484
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.870542988334809e-07,
+      "loss": 1.6459,
+      "step": 2485
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.862691970140134e-07,
+      "loss": 1.6622,
+      "step": 2486
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.854843096265919e-07,
+      "loss": 1.6438,
+      "step": 2487
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.846996372076785e-07,
+      "loss": 1.6651,
+      "step": 2488
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.839151802935903e-07,
+      "loss": 1.6464,
+      "step": 2489
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.831309394204956e-07,
+      "loss": 1.665,
+      "step": 2490
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.823469151244161e-07,
+      "loss": 1.6686,
+      "step": 2491
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.815631079412248e-07,
+      "loss": 1.6963,
+      "step": 2492
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.807795184066465e-07,
+      "loss": 1.6419,
+      "step": 2493
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.799961470562572e-07,
+      "loss": 1.658,
+      "step": 2494
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.792129944254841e-07,
+      "loss": 1.6337,
+      "step": 2495
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.784300610496047e-07,
+      "loss": 1.657,
+      "step": 2496
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.776473474637461e-07,
+      "loss": 1.6781,
+      "step": 2497
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.768648542028856e-07,
+      "loss": 1.6489,
+      "step": 2498
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.760825818018507e-07,
+      "loss": 1.6584,
+      "step": 2499
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.753005307953165e-07,
+      "loss": 1.6872,
+      "step": 2500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.74518701717808e-07,
+      "loss": 1.6477,
+      "step": 2501
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.737370951036976e-07,
+      "loss": 1.6273,
+      "step": 2502
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.729557114872063e-07,
+      "loss": 1.6632,
+      "step": 2503
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.721745514024022e-07,
+      "loss": 1.6662,
+      "step": 2504
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.713936153832012e-07,
+      "loss": 1.6638,
+      "step": 2505
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.706129039633658e-07,
+      "loss": 1.6415,
+      "step": 2506
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.698324176765041e-07,
+      "loss": 1.653,
+      "step": 2507
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.690521570560716e-07,
+      "loss": 1.6486,
+      "step": 2508
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.682721226353695e-07,
+      "loss": 1.6572,
+      "step": 2509
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.674923149475432e-07,
+      "loss": 1.6531,
+      "step": 2510
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.667127345255845e-07,
+      "loss": 1.6378,
+      "step": 2511
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.65933381902329e-07,
+      "loss": 1.662,
+      "step": 2512
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.651542576104568e-07,
+      "loss": 1.6549,
+      "step": 2513
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.643753621824921e-07,
+      "loss": 1.6417,
+      "step": 2514
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.635966961508031e-07,
+      "loss": 1.6444,
+      "step": 2515
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.628182600475997e-07,
+      "loss": 1.6665,
+      "step": 2516
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.620400544049363e-07,
+      "loss": 1.6728,
+      "step": 2517
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.612620797547086e-07,
+      "loss": 1.6633,
+      "step": 2518
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.604843366286557e-07,
+      "loss": 1.6812,
+      "step": 2519
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.597068255583569e-07,
+      "loss": 1.6758,
+      "step": 2520
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.589295470752343e-07,
+      "loss": 1.6634,
+      "step": 2521
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.581525017105501e-07,
+      "loss": 1.6714,
+      "step": 2522
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.573756899954077e-07,
+      "loss": 1.647,
+      "step": 2523
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.565991124607506e-07,
+      "loss": 1.6848,
+      "step": 2524
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.558227696373616e-07,
+      "loss": 1.6607,
+      "step": 2525
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.550466620558642e-07,
+      "loss": 1.6626,
+      "step": 2526
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.542707902467208e-07,
+      "loss": 1.6722,
+      "step": 2527
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.534951547402321e-07,
+      "loss": 1.682,
+      "step": 2528
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.527197560665378e-07,
+      "loss": 1.6498,
+      "step": 2529
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.519445947556155e-07,
+      "loss": 1.6764,
+      "step": 2530
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.51169671337281e-07,
+      "loss": 1.6832,
+      "step": 2531
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.503949863411865e-07,
+      "loss": 1.6438,
+      "step": 2532
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.496205402968224e-07,
+      "loss": 1.6459,
+      "step": 2533
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.488463337335156e-07,
+      "loss": 1.635,
+      "step": 2534
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.48072367180428e-07,
+      "loss": 1.6497,
+      "step": 2535
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.472986411665588e-07,
+      "loss": 1.6423,
+      "step": 2536
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.46525156220743e-07,
+      "loss": 1.642,
+      "step": 2537
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.457519128716495e-07,
+      "loss": 1.6659,
+      "step": 2538
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.449789116477835e-07,
+      "loss": 1.6346,
+      "step": 2539
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.442061530774834e-07,
+      "loss": 1.6597,
+      "step": 2540
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.434336376889231e-07,
+      "loss": 1.6691,
+      "step": 2541
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.42661366010109e-07,
+      "loss": 1.6481,
+      "step": 2542
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.41889338568882e-07,
+      "loss": 1.6621,
+      "step": 2543
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.411175558929152e-07,
+      "loss": 1.6677,
+      "step": 2544
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.403460185097147e-07,
+      "loss": 1.6577,
+      "step": 2545
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.395747269466191e-07,
+      "loss": 1.6386,
+      "step": 2546
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.388036817307993e-07,
+      "loss": 1.6411,
+      "step": 2547
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.38032883389257e-07,
+      "loss": 1.6476,
+      "step": 2548
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.37262332448826e-07,
+      "loss": 1.6376,
+      "step": 2549
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.364920294361699e-07,
+      "loss": 1.6436,
+      "step": 2550
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.357219748777846e-07,
+      "loss": 1.6707,
+      "step": 2551
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.349521692999944e-07,
+      "loss": 1.6524,
+      "step": 2552
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.341826132289541e-07,
+      "loss": 1.6283,
+      "step": 2553
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.334133071906482e-07,
+      "loss": 1.6414,
+      "step": 2554
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.326442517108901e-07,
+      "loss": 1.6555,
+      "step": 2555
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.31875447315322e-07,
+      "loss": 1.6578,
+      "step": 2556
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.311068945294147e-07,
+      "loss": 1.6628,
+      "step": 2557
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.303385938784662e-07,
+      "loss": 1.6493,
+      "step": 2558
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.295705458876032e-07,
+      "loss": 1.6503,
+      "step": 2559
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.288027510817791e-07,
+      "loss": 1.6669,
+      "step": 2560
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.280352099857746e-07,
+      "loss": 1.6723,
+      "step": 2561
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.27267923124197e-07,
+      "loss": 1.6534,
+      "step": 2562
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.265008910214786e-07,
+      "loss": 1.6717,
+      "step": 2563
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.257341142018797e-07,
+      "loss": 1.6288,
+      "step": 2564
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.249675931894845e-07,
+      "loss": 1.6523,
+      "step": 2565
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.242013285082029e-07,
+      "loss": 1.6537,
+      "step": 2566
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.2343532068177e-07,
+      "loss": 1.648,
+      "step": 2567
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.226695702337442e-07,
+      "loss": 1.6552,
+      "step": 2568
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.219040776875095e-07,
+      "loss": 1.6691,
+      "step": 2569
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.211388435662721e-07,
+      "loss": 1.6611,
+      "step": 2570
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.203738683930633e-07,
+      "loss": 1.6723,
+      "step": 2571
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.196091526907355e-07,
+      "loss": 1.6522,
+      "step": 2572
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.188446969819647e-07,
+      "loss": 1.6545,
+      "step": 2573
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.180805017892498e-07,
+      "loss": 1.6697,
+      "step": 2574
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.173165676349102e-07,
+      "loss": 1.6511,
+      "step": 2575
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.165528950410884e-07,
+      "loss": 1.6883,
+      "step": 2576
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.157894845297471e-07,
+      "loss": 1.6272,
+      "step": 2577
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.150263366226702e-07,
+      "loss": 1.656,
+      "step": 2578
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.142634518414622e-07,
+      "loss": 1.6414,
+      "step": 2579
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.13500830707548e-07,
+      "loss": 1.6582,
+      "step": 2580
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.12738473742171e-07,
+      "loss": 1.6395,
+      "step": 2581
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.119763814663957e-07,
+      "loss": 1.6739,
+      "step": 2582
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.112145544011047e-07,
+      "loss": 1.6579,
+      "step": 2583
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.10452993067e-07,
+      "loss": 1.6495,
+      "step": 2584
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.09691697984601e-07,
+      "loss": 1.6321,
+      "step": 2585
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.08930669674246e-07,
+      "loss": 1.6354,
+      "step": 2586
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.08169908656091e-07,
+      "loss": 1.645,
+      "step": 2587
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.074094154501086e-07,
+      "loss": 1.6762,
+      "step": 2588
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.066491905760891e-07,
+      "loss": 1.6502,
+      "step": 2589
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.058892345536386e-07,
+      "loss": 1.6417,
+      "step": 2590
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.051295479021798e-07,
+      "loss": 1.667,
+      "step": 2591
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.04370131140952e-07,
+      "loss": 1.6568,
+      "step": 2592
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.036109847890085e-07,
+      "loss": 1.6499,
+      "step": 2593
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.028521093652194e-07,
+      "loss": 1.6207,
+      "step": 2594
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.020935053882688e-07,
+      "loss": 1.6616,
+      "step": 2595
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.01335173376655e-07,
+      "loss": 1.6656,
+      "step": 2596
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.005771138486911e-07,
+      "loss": 1.6412,
+      "step": 2597
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.998193273225035e-07,
+      "loss": 1.6556,
+      "step": 2598
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.990618143160326e-07,
+      "loss": 1.6447,
+      "step": 2599
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.983045753470307e-07,
+      "loss": 1.675,
+      "step": 2600
+    },
+    {
+      "epoch": 1.83,
+      "eval_loss": 1.6577870845794678,
+      "eval_runtime": 41.4016,
+      "eval_samples_per_second": 48.307,
+      "eval_steps_per_second": 0.121,
+      "step": 2600
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.975476109330638e-07,
+      "loss": 1.6663,
+      "step": 2601
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.967909215915099e-07,
+      "loss": 1.7093,
+      "step": 2602
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.960345078395588e-07,
+      "loss": 1.6494,
+      "step": 2603
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.952783701942129e-07,
+      "loss": 1.6221,
+      "step": 2604
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.945225091722843e-07,
+      "loss": 1.6215,
+      "step": 2605
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.937669252903973e-07,
+      "loss": 1.6643,
+      "step": 2606
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.930116190649866e-07,
+      "loss": 1.6712,
+      "step": 2607
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.922565910122966e-07,
+      "loss": 1.6643,
+      "step": 2608
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.915018416483817e-07,
+      "loss": 1.6638,
+      "step": 2609
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.907473714891061e-07,
+      "loss": 1.6577,
+      "step": 2610
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.89993181050143e-07,
+      "loss": 1.6598,
+      "step": 2611
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.892392708469747e-07,
+      "loss": 1.6625,
+      "step": 2612
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.884856413948912e-07,
+      "loss": 1.6698,
+      "step": 2613
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.877322932089919e-07,
+      "loss": 1.6461,
+      "step": 2614
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.869792268041824e-07,
+      "loss": 1.6698,
+      "step": 2615
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.862264426951768e-07,
+      "loss": 1.6611,
+      "step": 2616
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.854739413964963e-07,
+      "loss": 1.6357,
+      "step": 2617
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.847217234224681e-07,
+      "loss": 1.6658,
+      "step": 2618
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.839697892872257e-07,
+      "loss": 1.6394,
+      "step": 2619
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.832181395047098e-07,
+      "loss": 1.6533,
+      "step": 2620
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.824667745886651e-07,
+      "loss": 1.6496,
+      "step": 2621
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.81715695052643e-07,
+      "loss": 1.6404,
+      "step": 2622
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.809649014099995e-07,
+      "loss": 1.6711,
+      "step": 2623
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.802143941738944e-07,
+      "loss": 1.6623,
+      "step": 2624
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.794641738572924e-07,
+      "loss": 1.6633,
+      "step": 2625
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.787142409729621e-07,
+      "loss": 1.6666,
+      "step": 2626
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.77964596033476e-07,
+      "loss": 1.6719,
+      "step": 2627
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.772152395512087e-07,
+      "loss": 1.6734,
+      "step": 2628
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.764661720383389e-07,
+      "loss": 1.6271,
+      "step": 2629
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.757173940068464e-07,
+      "loss": 1.6743,
+      "step": 2630
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.749689059685144e-07,
+      "loss": 1.6528,
+      "step": 2631
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.742207084349273e-07,
+      "loss": 1.6609,
+      "step": 2632
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.734728019174715e-07,
+      "loss": 1.6709,
+      "step": 2633
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.727251869273333e-07,
+      "loss": 1.6486,
+      "step": 2634
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.719778639755009e-07,
+      "loss": 1.6699,
+      "step": 2635
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.712308335727628e-07,
+      "loss": 1.659,
+      "step": 2636
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.704840962297066e-07,
+      "loss": 1.6603,
+      "step": 2637
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.697376524567204e-07,
+      "loss": 1.6653,
+      "step": 2638
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.689915027639922e-07,
+      "loss": 1.6336,
+      "step": 2639
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.682456476615072e-07,
+      "loss": 1.6458,
+      "step": 2640
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.675000876590508e-07,
+      "loss": 1.6558,
+      "step": 2641
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.667548232662063e-07,
+      "loss": 1.6394,
+      "step": 2642
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.660098549923552e-07,
+      "loss": 1.6534,
+      "step": 2643
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.652651833466755e-07,
+      "loss": 1.6735,
+      "step": 2644
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.645208088381441e-07,
+      "loss": 1.6818,
+      "step": 2645
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.637767319755332e-07,
+      "loss": 1.6582,
+      "step": 2646
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.630329532674126e-07,
+      "loss": 1.6876,
+      "step": 2647
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.622894732221482e-07,
+      "loss": 1.6476,
+      "step": 2648
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.615462923479018e-07,
+      "loss": 1.6613,
+      "step": 2649
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.608034111526297e-07,
+      "loss": 1.6294,
+      "step": 2650
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.600608301440848e-07,
+      "loss": 1.6363,
+      "step": 2651
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.593185498298141e-07,
+      "loss": 1.688,
+      "step": 2652
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.585765707171597e-07,
+      "loss": 1.6567,
+      "step": 2653
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.578348933132564e-07,
+      "loss": 1.6691,
+      "step": 2654
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.570935181250345e-07,
+      "loss": 1.6351,
+      "step": 2655
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.563524456592163e-07,
+      "loss": 1.6858,
+      "step": 2656
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.55611676422318e-07,
+      "loss": 1.6232,
+      "step": 2657
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.548712109206486e-07,
+      "loss": 1.6608,
+      "step": 2658
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.541310496603096e-07,
+      "loss": 1.6454,
+      "step": 2659
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.533911931471935e-07,
+      "loss": 1.6605,
+      "step": 2660
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.526516418869856e-07,
+      "loss": 1.6509,
+      "step": 2661
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.519123963851623e-07,
+      "loss": 1.6642,
+      "step": 2662
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.511734571469912e-07,
+      "loss": 1.6411,
+      "step": 2663
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.504348246775299e-07,
+      "loss": 1.6337,
+      "step": 2664
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.496964994816264e-07,
+      "loss": 1.6352,
+      "step": 2665
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.489584820639194e-07,
+      "loss": 1.6886,
+      "step": 2666
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.48220772928837e-07,
+      "loss": 1.666,
+      "step": 2667
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.474833725805962e-07,
+      "loss": 1.6526,
+      "step": 2668
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.467462815232035e-07,
+      "loss": 1.6376,
+      "step": 2669
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.460095002604532e-07,
+      "loss": 1.6612,
+      "step": 2670
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.452730292959286e-07,
+      "loss": 1.656,
+      "step": 2671
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.445368691330006e-07,
+      "loss": 1.6156,
+      "step": 2672
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.438010202748282e-07,
+      "loss": 1.6292,
+      "step": 2673
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.430654832243569e-07,
+      "loss": 1.6689,
+      "step": 2674
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.423302584843186e-07,
+      "loss": 1.6642,
+      "step": 2675
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.415953465572332e-07,
+      "loss": 1.6426,
+      "step": 2676
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.408607479454057e-07,
+      "loss": 1.66,
+      "step": 2677
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.401264631509277e-07,
+      "loss": 1.654,
+      "step": 2678
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.393924926756761e-07,
+      "loss": 1.6523,
+      "step": 2679
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.386588370213123e-07,
+      "loss": 1.6697,
+      "step": 2680
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.37925496689283e-07,
+      "loss": 1.6827,
+      "step": 2681
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.371924721808196e-07,
+      "loss": 1.6542,
+      "step": 2682
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.36459763996938e-07,
+      "loss": 1.6521,
+      "step": 2683
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.357273726384367e-07,
+      "loss": 1.6545,
+      "step": 2684
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.349952986058981e-07,
+      "loss": 1.6477,
+      "step": 2685
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.34263542399688e-07,
+      "loss": 1.637,
+      "step": 2686
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.33532104519955e-07,
+      "loss": 1.6621,
+      "step": 2687
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.328009854666302e-07,
+      "loss": 1.6555,
+      "step": 2688
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.320701857394267e-07,
+      "loss": 1.6483,
+      "step": 2689
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.313397058378386e-07,
+      "loss": 1.6591,
+      "step": 2690
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.306095462611424e-07,
+      "loss": 1.6758,
+      "step": 2691
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.298797075083955e-07,
+      "loss": 1.6521,
+      "step": 2692
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.291501900784352e-07,
+      "loss": 1.6626,
+      "step": 2693
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.284209944698805e-07,
+      "loss": 1.6309,
+      "step": 2694
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.276921211811292e-07,
+      "loss": 1.6491,
+      "step": 2695
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.269635707103593e-07,
+      "loss": 1.6558,
+      "step": 2696
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.262353435555283e-07,
+      "loss": 1.6438,
+      "step": 2697
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.255074402143727e-07,
+      "loss": 1.6447,
+      "step": 2698
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.247798611844079e-07,
+      "loss": 1.6644,
+      "step": 2699
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.240526069629264e-07,
+      "loss": 1.6379,
+      "step": 2700
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.233256780470007e-07,
+      "loss": 1.6297,
+      "step": 2701
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.225990749334788e-07,
+      "loss": 1.6678,
+      "step": 2702
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.218727981189876e-07,
+      "loss": 1.6382,
+      "step": 2703
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.211468480999304e-07,
+      "loss": 1.6816,
+      "step": 2704
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.204212253724876e-07,
+      "loss": 1.6182,
+      "step": 2705
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.196959304326147e-07,
+      "loss": 1.6287,
+      "step": 2706
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.189709637760444e-07,
+      "loss": 1.6656,
+      "step": 2707
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.182463258982846e-07,
+      "loss": 1.6563,
+      "step": 2708
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.175220172946187e-07,
+      "loss": 1.6502,
+      "step": 2709
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.16798038460104e-07,
+      "loss": 1.6282,
+      "step": 2710
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.160743898895741e-07,
+      "loss": 1.6122,
+      "step": 2711
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.153510720776353e-07,
+      "loss": 1.6541,
+      "step": 2712
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.146280855186687e-07,
+      "loss": 1.6381,
+      "step": 2713
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.139054307068287e-07,
+      "loss": 1.6776,
+      "step": 2714
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.131831081360438e-07,
+      "loss": 1.6614,
+      "step": 2715
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.124611183000137e-07,
+      "loss": 1.6208,
+      "step": 2716
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.11739461692212e-07,
+      "loss": 1.6351,
+      "step": 2717
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.110181388058842e-07,
+      "loss": 1.6532,
+      "step": 2718
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.102971501340481e-07,
+      "loss": 1.6528,
+      "step": 2719
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.095764961694922e-07,
+      "loss": 1.6615,
+      "step": 2720
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.088561774047764e-07,
+      "loss": 1.6633,
+      "step": 2721
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.081361943322322e-07,
+      "loss": 1.6618,
+      "step": 2722
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.074165474439611e-07,
+      "loss": 1.6682,
+      "step": 2723
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.06697237231835e-07,
+      "loss": 1.5972,
+      "step": 2724
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.059782641874961e-07,
+      "loss": 1.6582,
+      "step": 2725
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.05259628802355e-07,
+      "loss": 1.6572,
+      "step": 2726
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.045413315675923e-07,
+      "loss": 1.6458,
+      "step": 2727
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.038233729741577e-07,
+      "loss": 1.6454,
+      "step": 2728
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.031057535127694e-07,
+      "loss": 1.6478,
+      "step": 2729
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.023884736739131e-07,
+      "loss": 1.6455,
+      "step": 2730
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.016715339478425e-07,
+      "loss": 1.6611,
+      "step": 2731
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.009549348245795e-07,
+      "loss": 1.6661,
+      "step": 2732
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.002386767939129e-07,
+      "loss": 1.6146,
+      "step": 2733
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.99522760345398e-07,
+      "loss": 1.6727,
+      "step": 2734
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.988071859683578e-07,
+      "loss": 1.6698,
+      "step": 2735
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.980919541518795e-07,
+      "loss": 1.6497,
+      "step": 2736
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.97377065384818e-07,
+      "loss": 1.6598,
+      "step": 2737
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.96662520155793e-07,
+      "loss": 1.6667,
+      "step": 2738
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.959483189531891e-07,
+      "loss": 1.65,
+      "step": 2739
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.952344622651566e-07,
+      "loss": 1.6474,
+      "step": 2740
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.94520950579609e-07,
+      "loss": 1.6456,
+      "step": 2741
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.938077843842255e-07,
+      "loss": 1.6445,
+      "step": 2742
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.930949641664482e-07,
+      "loss": 1.6441,
+      "step": 2743
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.923824904134829e-07,
+      "loss": 1.6487,
+      "step": 2744
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.916703636122994e-07,
+      "loss": 1.6712,
+      "step": 2745
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.909585842496287e-07,
+      "loss": 1.6674,
+      "step": 2746
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.902471528119656e-07,
+      "loss": 1.6336,
+      "step": 2747
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.895360697855674e-07,
+      "loss": 1.6583,
+      "step": 2748
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.888253356564517e-07,
+      "loss": 1.6371,
+      "step": 2749
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.881149509103992e-07,
+      "loss": 1.6556,
+      "step": 2750
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.874049160329508e-07,
+      "loss": 1.6655,
+      "step": 2751
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.866952315094087e-07,
+      "loss": 1.661,
+      "step": 2752
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.859858978248357e-07,
+      "loss": 1.6576,
+      "step": 2753
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.852769154640547e-07,
+      "loss": 1.6579,
+      "step": 2754
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.845682849116488e-07,
+      "loss": 1.6338,
+      "step": 2755
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.838600066519596e-07,
+      "loss": 1.6429,
+      "step": 2756
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.831520811690893e-07,
+      "loss": 1.6605,
+      "step": 2757
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.824445089468975e-07,
+      "loss": 1.6518,
+      "step": 2758
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.817372904690036e-07,
+      "loss": 1.6299,
+      "step": 2759
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.810304262187851e-07,
+      "loss": 1.6255,
+      "step": 2760
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.803239166793763e-07,
+      "loss": 1.657,
+      "step": 2761
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.796177623336702e-07,
+      "loss": 1.6532,
+      "step": 2762
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.789119636643168e-07,
+      "loss": 1.6751,
+      "step": 2763
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.782065211537225e-07,
+      "loss": 1.6658,
+      "step": 2764
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.775014352840512e-07,
+      "loss": 1.6597,
+      "step": 2765
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.76796706537222e-07,
+      "loss": 1.641,
+      "step": 2766
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.7609233539491e-07,
+      "loss": 1.6529,
+      "step": 2767
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.7538832233854665e-07,
+      "loss": 1.6472,
+      "step": 2768
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.7468466784931803e-07,
+      "loss": 1.6754,
+      "step": 2769
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.7398137240816603e-07,
+      "loss": 1.6362,
+      "step": 2770
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.732784364957856e-07,
+      "loss": 1.622,
+      "step": 2771
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.72575860592627e-07,
+      "loss": 1.6626,
+      "step": 2772
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.718736451788944e-07,
+      "loss": 1.6669,
+      "step": 2773
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.711717907345455e-07,
+      "loss": 1.6505,
+      "step": 2774
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.704702977392914e-07,
+      "loss": 1.6113,
+      "step": 2775
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.6976916667259545e-07,
+      "loss": 1.6615,
+      "step": 2776
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.6906839801367415e-07,
+      "loss": 1.6461,
+      "step": 2777
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.6836799224149637e-07,
+      "loss": 1.6678,
+      "step": 2778
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.6766794983478287e-07,
+      "loss": 1.6279,
+      "step": 2779
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.6696827127200645e-07,
+      "loss": 1.6739,
+      "step": 2780
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.6626895703139e-07,
+      "loss": 1.6524,
+      "step": 2781
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.655700075909087e-07,
+      "loss": 1.6472,
+      "step": 2782
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.648714234282879e-07,
+      "loss": 1.6506,
+      "step": 2783
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.641732050210031e-07,
+      "loss": 1.6413,
+      "step": 2784
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.6347535284628056e-07,
+      "loss": 1.6446,
+      "step": 2785
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.627778673810953e-07,
+      "loss": 1.6457,
+      "step": 2786
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.6208074910217143e-07,
+      "loss": 1.6517,
+      "step": 2787
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.613839984859834e-07,
+      "loss": 1.6547,
+      "step": 2788
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.606876160087535e-07,
+      "loss": 1.6419,
+      "step": 2789
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.599916021464531e-07,
+      "loss": 1.6372,
+      "step": 2790
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.592959573748002e-07,
+      "loss": 1.6757,
+      "step": 2791
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.5860068216926193e-07,
+      "loss": 1.6697,
+      "step": 2792
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.579057770050524e-07,
+      "loss": 1.6798,
+      "step": 2793
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.572112423571329e-07,
+      "loss": 1.6489,
+      "step": 2794
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.5651707870021075e-07,
+      "loss": 1.6591,
+      "step": 2795
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.5582328650874093e-07,
+      "loss": 1.6431,
+      "step": 2796
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.5512986625692306e-07,
+      "loss": 1.6434,
+      "step": 2797
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.5443681841870383e-07,
+      "loss": 1.6562,
+      "step": 2798
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.5374414346777477e-07,
+      "loss": 1.6419,
+      "step": 2799
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.530518418775733e-07,
+      "loss": 1.6546,
+      "step": 2800
+    },
+    {
+      "epoch": 1.97,
+      "eval_loss": 1.6561248302459717,
+      "eval_runtime": 42.5741,
+      "eval_samples_per_second": 46.977,
+      "eval_steps_per_second": 0.117,
+      "step": 2800
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.5235991412128004e-07,
+      "loss": 1.673,
+      "step": 2801
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.516683606718216e-07,
+      "loss": 1.6559,
+      "step": 2802
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.5097718200186805e-07,
+      "loss": 1.6341,
+      "step": 2803
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.502863785838341e-07,
+      "loss": 1.64,
+      "step": 2804
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.4959595088987654e-07,
+      "loss": 1.6815,
+      "step": 2805
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.489058993918968e-07,
+      "loss": 1.6293,
+      "step": 2806
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.4821622456153804e-07,
+      "loss": 1.655,
+      "step": 2807
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.475269268701868e-07,
+      "loss": 1.6215,
+      "step": 2808
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.4683800678897155e-07,
+      "loss": 1.643,
+      "step": 2809
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.4614946478876303e-07,
+      "loss": 1.6407,
+      "step": 2810
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.454613013401723e-07,
+      "loss": 1.649,
+      "step": 2811
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.447735169135532e-07,
+      "loss": 1.6403,
+      "step": 2812
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.4408611197900023e-07,
+      "loss": 1.6398,
+      "step": 2813
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.4339908700634744e-07,
+      "loss": 1.6548,
+      "step": 2814
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.4271244246517026e-07,
+      "loss": 1.6737,
+      "step": 2815
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.42026178824784e-07,
+      "loss": 1.6684,
+      "step": 2816
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.4134029655424274e-07,
+      "loss": 1.6455,
+      "step": 2817
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.4065479612234115e-07,
+      "loss": 1.6364,
+      "step": 2818
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.399696779976119e-07,
+      "loss": 1.6257,
+      "step": 2819
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.3928494264832736e-07,
+      "loss": 1.6587,
+      "step": 2820
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.3860059054249696e-07,
+      "loss": 1.6575,
+      "step": 2821
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.3791662214786963e-07,
+      "loss": 1.6624,
+      "step": 2822
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.3723303793193054e-07,
+      "loss": 1.6448,
+      "step": 2823
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.3654983836190353e-07,
+      "loss": 1.6535,
+      "step": 2824
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.358670239047489e-07,
+      "loss": 1.6462,
+      "step": 2825
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.351845950271643e-07,
+      "loss": 1.6387,
+      "step": 2826
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.3450255219558284e-07,
+      "loss": 1.6656,
+      "step": 2827
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.3382089587617466e-07,
+      "loss": 1.6369,
+      "step": 2828
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.3313962653484524e-07,
+      "loss": 1.6829,
+      "step": 2829
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.3245874463723643e-07,
+      "loss": 1.6419,
+      "step": 2830
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.3177825064872366e-07,
+      "loss": 1.6916,
+      "step": 2831
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.310981450344189e-07,
+      "loss": 1.6297,
+      "step": 2832
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.304184282591672e-07,
+      "loss": 1.642,
+      "step": 2833
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.2973910078754895e-07,
+      "loss": 1.6265,
+      "step": 2834
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.2906016308387803e-07,
+      "loss": 1.6392,
+      "step": 2835
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.2838161561220244e-07,
+      "loss": 1.6346,
+      "step": 2836
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.277034588363022e-07,
+      "loss": 1.6395,
+      "step": 2837
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.270256932196914e-07,
+      "loss": 1.6458,
+      "step": 2838
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.2634831922561643e-07,
+      "loss": 1.669,
+      "step": 2839
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.256713373170564e-07,
+      "loss": 1.6864,
+      "step": 2840
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.2499474795672177e-07,
+      "loss": 1.6727,
+      "step": 2841
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.2431855160705455e-07,
+      "loss": 1.6713,
+      "step": 2842
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.2364274873022897e-07,
+      "loss": 1.6235,
+      "step": 2843
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.2296733978814993e-07,
+      "loss": 1.6678,
+      "step": 2844
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.2229232524245293e-07,
+      "loss": 1.6442,
+      "step": 2845
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.216177055545047e-07,
+      "loss": 1.6502,
+      "step": 2846
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.2094348118540046e-07,
+      "loss": 1.6402,
+      "step": 2847
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.202696525959666e-07,
+      "loss": 1.6511,
+      "step": 2848
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.195962202467588e-07,
+      "loss": 1.6339,
+      "step": 2849
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.1892318459806175e-07,
+      "loss": 1.6561,
+      "step": 2850
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.1825054610988897e-07,
+      "loss": 1.6639,
+      "step": 2851
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.175783052419819e-07,
+      "loss": 1.6621,
+      "step": 2852
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.169064624538113e-07,
+      "loss": 1.6609,
+      "step": 2853
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.162350182045754e-07,
+      "loss": 1.6305,
+      "step": 2854
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.155639729532e-07,
+      "loss": 1.6533,
+      "step": 2855
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.148933271583385e-07,
+      "loss": 1.6578,
+      "step": 2856
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.1422308127837036e-07,
+      "loss": 1.6436,
+      "step": 2857
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.1355323577140257e-07,
+      "loss": 1.6516,
+      "step": 2858
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.128837910952683e-07,
+      "loss": 1.6658,
+      "step": 2859
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.1221474770752696e-07,
+      "loss": 1.6735,
+      "step": 2860
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.11546106065463e-07,
+      "loss": 1.6355,
+      "step": 2861
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.108778666260866e-07,
+      "loss": 1.6576,
+      "step": 2862
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.1021002984613305e-07,
+      "loss": 1.6861,
+      "step": 2863
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.095425961820629e-07,
+      "loss": 1.67,
+      "step": 2864
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.0887556609006046e-07,
+      "loss": 1.6719,
+      "step": 2865
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.0820894002603503e-07,
+      "loss": 1.6516,
+      "step": 2866
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.075427184456185e-07,
+      "loss": 1.6583,
+      "step": 2867
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.0687690180416735e-07,
+      "loss": 1.6589,
+      "step": 2868
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.062114905567614e-07,
+      "loss": 1.6461,
+      "step": 2869
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.0554648515820213e-07,
+      "loss": 1.6442,
+      "step": 2870
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.048818860630153e-07,
+      "loss": 1.6377,
+      "step": 2871
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.0421769372544735e-07,
+      "loss": 1.6594,
+      "step": 2872
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.035539085994678e-07,
+      "loss": 1.6596,
+      "step": 2873
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.028905311387677e-07,
+      "loss": 1.6608,
+      "step": 2874
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.022275617967591e-07,
+      "loss": 1.646,
+      "step": 2875
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.0156500102657565e-07,
+      "loss": 1.6401,
+      "step": 2876
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.0090284928107074e-07,
+      "loss": 1.6614,
+      "step": 2877
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.0024110701281967e-07,
+      "loss": 1.651,
+      "step": 2878
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.9957977467411605e-07,
+      "loss": 1.6213,
+      "step": 2879
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.989188527169749e-07,
+      "loss": 1.6451,
+      "step": 2880
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.982583415931304e-07,
+      "loss": 1.6193,
+      "step": 2881
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.97598241754035e-07,
+      "loss": 1.6378,
+      "step": 2882
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.96938553650861e-07,
+      "loss": 1.6671,
+      "step": 2883
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.962792777344992e-07,
+      "loss": 1.6623,
+      "step": 2884
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9562041445555815e-07,
+      "loss": 1.6523,
+      "step": 2885
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9496196426436535e-07,
+      "loss": 1.6547,
+      "step": 2886
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.943039276109644e-07,
+      "loss": 1.6687,
+      "step": 2887
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9364630494511785e-07,
+      "loss": 1.6085,
+      "step": 2888
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.92989096716304e-07,
+      "loss": 1.6171,
+      "step": 2889
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9233230337371883e-07,
+      "loss": 1.6436,
+      "step": 2890
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.916759253662746e-07,
+      "loss": 1.6222,
+      "step": 2891
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.910199631425989e-07,
+      "loss": 1.6501,
+      "step": 2892
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.903644171510361e-07,
+      "loss": 1.6854,
+      "step": 2893
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.897092878396456e-07,
+      "loss": 1.6553,
+      "step": 2894
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.890545756562021e-07,
+      "loss": 1.6558,
+      "step": 2895
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.884002810481958e-07,
+      "loss": 1.6668,
+      "step": 2896
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.8774640446283023e-07,
+      "loss": 1.6383,
+      "step": 2897
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.8709294634702373e-07,
+      "loss": 1.6292,
+      "step": 2898
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.864399071474089e-07,
+      "loss": 1.657,
+      "step": 2899
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.8578728731033214e-07,
+      "loss": 1.6609,
+      "step": 2900
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.8513508728185316e-07,
+      "loss": 1.6398,
+      "step": 2901
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.8448330750774383e-07,
+      "loss": 1.6307,
+      "step": 2902
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.8383194843348975e-07,
+      "loss": 1.6337,
+      "step": 2903
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.83181010504289e-07,
+      "loss": 1.6442,
+      "step": 2904
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.825304941650511e-07,
+      "loss": 1.6392,
+      "step": 2905
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.8188039986039854e-07,
+      "loss": 1.6579,
+      "step": 2906
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.8123072803466406e-07,
+      "loss": 1.6439,
+      "step": 2907
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.805814791318921e-07,
+      "loss": 1.66,
+      "step": 2908
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.7993265359583826e-07,
+      "loss": 1.6621,
+      "step": 2909
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.792842518699688e-07,
+      "loss": 1.6572,
+      "step": 2910
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.786362743974606e-07,
+      "loss": 1.6446,
+      "step": 2911
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.7798872162119944e-07,
+      "loss": 1.6554,
+      "step": 2912
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.7734159398378173e-07,
+      "loss": 1.6561,
+      "step": 2913
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.7669489192751323e-07,
+      "loss": 1.6712,
+      "step": 2914
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.760486158944091e-07,
+      "loss": 1.6205,
+      "step": 2915
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.754027663261922e-07,
+      "loss": 1.6706,
+      "step": 2916
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.747573436642951e-07,
+      "loss": 1.6533,
+      "step": 2917
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.741123483498576e-07,
+      "loss": 1.6419,
+      "step": 2918
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.7346778082372833e-07,
+      "loss": 1.66,
+      "step": 2919
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.7282364152646295e-07,
+      "loss": 1.6403,
+      "step": 2920
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.72179930898325e-07,
+      "loss": 1.6644,
+      "step": 2921
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.7153664937928396e-07,
+      "loss": 1.651,
+      "step": 2922
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.7089379740901703e-07,
+      "loss": 1.6348,
+      "step": 2923
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.7025137542690755e-07,
+      "loss": 1.6226,
+      "step": 2924
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.69609383872045e-07,
+      "loss": 1.642,
+      "step": 2925
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.6896782318322407e-07,
+      "loss": 1.6704,
+      "step": 2926
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.6832669379894597e-07,
+      "loss": 1.6497,
+      "step": 2927
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.676859961574161e-07,
+      "loss": 1.6471,
+      "step": 2928
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.6704573069654544e-07,
+      "loss": 1.6594,
+      "step": 2929
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.664058978539495e-07,
+      "loss": 1.6638,
+      "step": 2930
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.657664980669478e-07,
+      "loss": 1.6621,
+      "step": 2931
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.651275317725647e-07,
+      "loss": 1.673,
+      "step": 2932
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.6448899940752674e-07,
+      "loss": 1.6518,
+      "step": 2933
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.638509014082656e-07,
+      "loss": 1.6454,
+      "step": 2934
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.632132382109143e-07,
+      "loss": 1.626,
+      "step": 2935
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.625760102513102e-07,
+      "loss": 1.6632,
+      "step": 2936
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.619392179649931e-07,
+      "loss": 1.6549,
+      "step": 2937
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.613028617872034e-07,
+      "loss": 1.6308,
+      "step": 2938
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.606669421528852e-07,
+      "loss": 1.6606,
+      "step": 2939
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.6003145949668337e-07,
+      "loss": 1.6421,
+      "step": 2940
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.5939641425294454e-07,
+      "loss": 1.6418,
+      "step": 2941
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.587618068557163e-07,
+      "loss": 1.6326,
+      "step": 2942
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.5812763773874666e-07,
+      "loss": 1.6388,
+      "step": 2943
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.574939073354838e-07,
+      "loss": 1.6516,
+      "step": 2944
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.568606160790767e-07,
+      "loss": 1.6463,
+      "step": 2945
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.562277644023741e-07,
+      "loss": 1.6418,
+      "step": 2946
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.555953527379245e-07,
+      "loss": 1.6336,
+      "step": 2947
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.5496338151797455e-07,
+      "loss": 1.6437,
+      "step": 2948
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.5433185117447107e-07,
+      "loss": 1.6328,
+      "step": 2949
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.53700762139059e-07,
+      "loss": 1.6627,
+      "step": 2950
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.530701148430818e-07,
+      "loss": 1.6459,
+      "step": 2951
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.524399097175812e-07,
+      "loss": 1.6674,
+      "step": 2952
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.518101471932962e-07,
+      "loss": 1.6312,
+      "step": 2953
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.5118082770066315e-07,
+      "loss": 1.6509,
+      "step": 2954
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.5055195166981646e-07,
+      "loss": 1.6536,
+      "step": 2955
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.4992351953058674e-07,
+      "loss": 1.632,
+      "step": 2956
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.49295531712502e-07,
+      "loss": 1.6642,
+      "step": 2957
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.486679886447852e-07,
+      "loss": 1.6682,
+      "step": 2958
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.4804089075635647e-07,
+      "loss": 1.6315,
+      "step": 2959
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.4741423847583127e-07,
+      "loss": 1.6439,
+      "step": 2960
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.4678803223152063e-07,
+      "loss": 1.6457,
+      "step": 2961
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.461622724514308e-07,
+      "loss": 1.6615,
+      "step": 2962
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.4553695956326245e-07,
+      "loss": 1.6595,
+      "step": 2963
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.449120939944107e-07,
+      "loss": 1.6664,
+      "step": 2964
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.4428767617196564e-07,
+      "loss": 1.6494,
+      "step": 2965
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.436637065227108e-07,
+      "loss": 1.6695,
+      "step": 2966
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.430401854731241e-07,
+      "loss": 1.6386,
+      "step": 2967
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.424171134493755e-07,
+      "loss": 1.6367,
+      "step": 2968
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.417944908773293e-07,
+      "loss": 1.6278,
+      "step": 2969
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.41172318182542e-07,
+      "loss": 1.6574,
+      "step": 2970
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.405505957902631e-07,
+      "loss": 1.6637,
+      "step": 2971
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.399293241254335e-07,
+      "loss": 1.6245,
+      "step": 2972
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.3930850361268716e-07,
+      "loss": 1.639,
+      "step": 2973
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.3868813467634827e-07,
+      "loss": 1.6547,
+      "step": 2974
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.380682177404335e-07,
+      "loss": 1.6774,
+      "step": 2975
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.374487532286503e-07,
+      "loss": 1.6424,
+      "step": 2976
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.3682974156439705e-07,
+      "loss": 1.6353,
+      "step": 2977
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.362111831707618e-07,
+      "loss": 1.6538,
+      "step": 2978
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.355930784705238e-07,
+      "loss": 1.6454,
+      "step": 2979
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.349754278861516e-07,
+      "loss": 1.6535,
+      "step": 2980
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.343582318398039e-07,
+      "loss": 1.6454,
+      "step": 2981
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.337414907533279e-07,
+      "loss": 1.6642,
+      "step": 2982
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.331252050482608e-07,
+      "loss": 1.638,
+      "step": 2983
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.3250937514582753e-07,
+      "loss": 1.6329,
+      "step": 2984
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.3189400146694226e-07,
+      "loss": 1.6492,
+      "step": 2985
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.312790844322071e-07,
+      "loss": 1.6526,
+      "step": 2986
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.3066462446191233e-07,
+      "loss": 1.68,
+      "step": 2987
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.3005062197603506e-07,
+      "loss": 1.6387,
+      "step": 2988
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.294370773942404e-07,
+      "loss": 1.6521,
+      "step": 2989
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.288239911358807e-07,
+      "loss": 1.6462,
+      "step": 2990
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.282113636199938e-07,
+      "loss": 1.6481,
+      "step": 2991
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.275991952653053e-07,
+      "loss": 1.6606,
+      "step": 2992
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.269874864902269e-07,
+      "loss": 1.6292,
+      "step": 2993
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2637623771285493e-07,
+      "loss": 1.6368,
+      "step": 2994
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.257654493509726e-07,
+      "loss": 1.6217,
+      "step": 2995
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.25155121822048e-07,
+      "loss": 1.6264,
+      "step": 2996
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2454525554323455e-07,
+      "loss": 1.664,
+      "step": 2997
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2393585093136943e-07,
+      "loss": 1.6543,
+      "step": 2998
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2332690840297574e-07,
+      "loss": 1.6611,
+      "step": 2999
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.227184283742591e-07,
+      "loss": 1.6373,
+      "step": 3000
+    },
+    {
+      "epoch": 2.11,
+      "eval_loss": 1.6550053358078003,
+      "eval_runtime": 42.7689,
+      "eval_samples_per_second": 46.763,
+      "eval_steps_per_second": 0.117,
+      "step": 3000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2211041126111027e-07,
+      "loss": 1.6497,
+      "step": 3001
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.215028574791032e-07,
+      "loss": 1.6363,
+      "step": 3002
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2089576744349556e-07,
+      "loss": 1.6726,
+      "step": 3003
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.20289141569227e-07,
+      "loss": 1.6637,
+      "step": 3004
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.1968298027092087e-07,
+      "loss": 1.6341,
+      "step": 3005
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.190772839628827e-07,
+      "loss": 1.6379,
+      "step": 3006
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.184720530591005e-07,
+      "loss": 1.6457,
+      "step": 3007
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.178672879732435e-07,
+      "loss": 1.6479,
+      "step": 3008
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.1726298911866314e-07,
+      "loss": 1.6481,
+      "step": 3009
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.1665915690839163e-07,
+      "loss": 1.6544,
+      "step": 3010
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.1605579175514286e-07,
+      "loss": 1.6518,
+      "step": 3011
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.154528940713113e-07,
+      "loss": 1.6397,
+      "step": 3012
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.14850464268972e-07,
+      "loss": 1.6545,
+      "step": 3013
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.1424850275987947e-07,
+      "loss": 1.669,
+      "step": 3014
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.13647009955469e-07,
+      "loss": 1.6631,
+      "step": 3015
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.130459862668554e-07,
+      "loss": 1.6645,
+      "step": 3016
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.124454321048329e-07,
+      "loss": 1.6652,
+      "step": 3017
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.1184534787987425e-07,
+      "loss": 1.6491,
+      "step": 3018
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.1124573400213116e-07,
+      "loss": 1.6455,
+      "step": 3019
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.106465908814342e-07,
+      "loss": 1.649,
+      "step": 3020
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.100479189272923e-07,
+      "loss": 1.6423,
+      "step": 3021
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.094497185488919e-07,
+      "loss": 1.6805,
+      "step": 3022
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.0885199015509764e-07,
+      "loss": 1.6283,
+      "step": 3023
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.082547341544507e-07,
+      "loss": 1.6261,
+      "step": 3024
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.076579509551702e-07,
+      "loss": 1.6473,
+      "step": 3025
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.0706164096515187e-07,
+      "loss": 1.6446,
+      "step": 3026
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.0646580459196813e-07,
+      "loss": 1.6404,
+      "step": 3027
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.058704422428674e-07,
+      "loss": 1.6417,
+      "step": 3028
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.0527555432477383e-07,
+      "loss": 1.6694,
+      "step": 3029
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.04681141244288e-07,
+      "loss": 1.6491,
+      "step": 3030
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.040872034076857e-07,
+      "loss": 1.6307,
+      "step": 3031
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.034937412209178e-07,
+      "loss": 1.651,
+      "step": 3032
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.0290075508961035e-07,
+      "loss": 1.6085,
+      "step": 3033
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.023082454190633e-07,
+      "loss": 1.6226,
+      "step": 3034
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.017162126142516e-07,
+      "loss": 1.6263,
+      "step": 3035
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.0112465707982416e-07,
+      "loss": 1.6557,
+      "step": 3036
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.00533579220104e-07,
+      "loss": 1.6627,
+      "step": 3037
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.999429794390869e-07,
+      "loss": 1.6466,
+      "step": 3038
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.9935285814044175e-07,
+      "loss": 1.5995,
+      "step": 3039
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.987632157275114e-07,
+      "loss": 1.646,
+      "step": 3040
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.981740526033109e-07,
+      "loss": 1.6427,
+      "step": 3041
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.975853691705276e-07,
+      "loss": 1.6505,
+      "step": 3042
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.9699716583152135e-07,
+      "loss": 1.6325,
+      "step": 3043
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.9640944298832305e-07,
+      "loss": 1.6532,
+      "step": 3044
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.95822201042636e-07,
+      "loss": 1.6718,
+      "step": 3045
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.952354403958349e-07,
+      "loss": 1.6476,
+      "step": 3046
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.9464916144896445e-07,
+      "loss": 1.6327,
+      "step": 3047
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.940633646027414e-07,
+      "loss": 1.6455,
+      "step": 3048
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.9347805025755167e-07,
+      "loss": 1.6227,
+      "step": 3049
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.9289321881345254e-07,
+      "loss": 1.6317,
+      "step": 3050
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.9230887067017075e-07,
+      "loss": 1.6545,
+      "step": 3051
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.9172500622710263e-07,
+      "loss": 1.6322,
+      "step": 3052
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.9114162588331446e-07,
+      "loss": 1.6523,
+      "step": 3053
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.9055873003754063e-07,
+      "loss": 1.6502,
+      "step": 3054
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8997631908818555e-07,
+      "loss": 1.6397,
+      "step": 3055
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8939439343332085e-07,
+      "loss": 1.6247,
+      "step": 3056
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8881295347068755e-07,
+      "loss": 1.6454,
+      "step": 3057
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8823199959769486e-07,
+      "loss": 1.6833,
+      "step": 3058
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.876515322114186e-07,
+      "loss": 1.6571,
+      "step": 3059
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.87071551708603e-07,
+      "loss": 1.6763,
+      "step": 3060
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.864920584856593e-07,
+      "loss": 1.6476,
+      "step": 3061
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.859130529386655e-07,
+      "loss": 1.6355,
+      "step": 3062
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8533453546336706e-07,
+      "loss": 1.6443,
+      "step": 3063
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.847565064551747e-07,
+      "loss": 1.6187,
+      "step": 3064
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.841789663091655e-07,
+      "loss": 1.6507,
+      "step": 3065
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8360191542008306e-07,
+      "loss": 1.6309,
+      "step": 3066
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.8302535418233617e-07,
+      "loss": 1.6562,
+      "step": 3067
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.824492829899994e-07,
+      "loss": 1.6788,
+      "step": 3068
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.818737022368113e-07,
+      "loss": 1.6265,
+      "step": 3069
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.812986123161761e-07,
+      "loss": 1.6711,
+      "step": 3070
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.807240136211625e-07,
+      "loss": 1.6241,
+      "step": 3071
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.801499065445032e-07,
+      "loss": 1.6457,
+      "step": 3072
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.795762914785954e-07,
+      "loss": 1.6566,
+      "step": 3073
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.7900316881549913e-07,
+      "loss": 1.6444,
+      "step": 3074
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.7843053894693803e-07,
+      "loss": 1.6491,
+      "step": 3075
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.778584022642996e-07,
+      "loss": 1.6349,
+      "step": 3076
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.7728675915863377e-07,
+      "loss": 1.6502,
+      "step": 3077
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.767156100206536e-07,
+      "loss": 1.6035,
+      "step": 3078
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.7614495524073355e-07,
+      "loss": 1.6549,
+      "step": 3079
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.7557479520891104e-07,
+      "loss": 1.6509,
+      "step": 3080
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.750051303148851e-07,
+      "loss": 1.6448,
+      "step": 3081
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.7443596094801624e-07,
+      "loss": 1.6223,
+      "step": 3082
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.738672874973268e-07,
+      "loss": 1.6271,
+      "step": 3083
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.7329911035149934e-07,
+      "loss": 1.6416,
+      "step": 3084
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.727314298988772e-07,
+      "loss": 1.6507,
+      "step": 3085
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.7216424652746496e-07,
+      "loss": 1.6448,
+      "step": 3086
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.715975606249272e-07,
+      "loss": 1.6379,
+      "step": 3087
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.7103137257858863e-07,
+      "loss": 1.6493,
+      "step": 3088
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.704656827754328e-07,
+      "loss": 1.6283,
+      "step": 3089
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.699004916021038e-07,
+      "loss": 1.639,
+      "step": 3090
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.693357994449043e-07,
+      "loss": 1.6385,
+      "step": 3091
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.6877160668979636e-07,
+      "loss": 1.6666,
+      "step": 3092
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.6820791372239993e-07,
+      "loss": 1.6729,
+      "step": 3093
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.6764472092799437e-07,
+      "loss": 1.6269,
+      "step": 3094
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.670820286915161e-07,
+      "loss": 1.6395,
+      "step": 3095
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.6651983739756023e-07,
+      "loss": 1.6661,
+      "step": 3096
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.659581474303791e-07,
+      "loss": 1.6228,
+      "step": 3097
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.6539695917388296e-07,
+      "loss": 1.6273,
+      "step": 3098
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.6483627301163804e-07,
+      "loss": 1.6467,
+      "step": 3099
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.642760893268684e-07,
+      "loss": 1.6365,
+      "step": 3100
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.6371640850245415e-07,
+      "loss": 1.6305,
+      "step": 3101
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.631572309209322e-07,
+      "loss": 1.65,
+      "step": 3102
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.6259855696449475e-07,
+      "loss": 1.6379,
+      "step": 3103
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.6204038701499053e-07,
+      "loss": 1.6265,
+      "step": 3104
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.614827214539229e-07,
+      "loss": 1.6596,
+      "step": 3105
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.6092556066245153e-07,
+      "loss": 1.6371,
+      "step": 3106
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.6036890502139017e-07,
+      "loss": 1.6733,
+      "step": 3107
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.598127549112084e-07,
+      "loss": 1.6496,
+      "step": 3108
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.592571107120287e-07,
+      "loss": 1.6474,
+      "step": 3109
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5870197280362914e-07,
+      "loss": 1.6425,
+      "step": 3110
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.581473415654414e-07,
+      "loss": 1.6364,
+      "step": 3111
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.575932173765502e-07,
+      "loss": 1.6294,
+      "step": 3112
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.570396006156943e-07,
+      "loss": 1.6269,
+      "step": 3113
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.56486491661266e-07,
+      "loss": 1.6479,
+      "step": 3114
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5593389089130955e-07,
+      "loss": 1.6726,
+      "step": 3115
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.553817986835225e-07,
+      "loss": 1.6393,
+      "step": 3116
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.548302154152547e-07,
+      "loss": 1.6544,
+      "step": 3117
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.542791414635085e-07,
+      "loss": 1.6614,
+      "step": 3118
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5372857720493704e-07,
+      "loss": 1.6655,
+      "step": 3119
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5317852301584643e-07,
+      "loss": 1.6715,
+      "step": 3120
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5262897927219283e-07,
+      "loss": 1.6423,
+      "step": 3121
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.520799463495847e-07,
+      "loss": 1.6517,
+      "step": 3122
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5153142462328094e-07,
+      "loss": 1.6565,
+      "step": 3123
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.5098341446819093e-07,
+      "loss": 1.6769,
+      "step": 3124
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.5043591625887404e-07,
+      "loss": 1.6449,
+      "step": 3125
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.498889303695404e-07,
+      "loss": 1.6342,
+      "step": 3126
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.4934245717404976e-07,
+      "loss": 1.6358,
+      "step": 3127
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.487964970459118e-07,
+      "loss": 1.6475,
+      "step": 3128
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.482510503582842e-07,
+      "loss": 1.6549,
+      "step": 3129
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.477061174839755e-07,
+      "loss": 1.6645,
+      "step": 3130
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.4716169879544146e-07,
+      "loss": 1.6542,
+      "step": 3131
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.4661779466478736e-07,
+      "loss": 1.6304,
+      "step": 3132
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.4607440546376676e-07,
+      "loss": 1.6518,
+      "step": 3133
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.4553153156378126e-07,
+      "loss": 1.6826,
+      "step": 3134
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.449891733358793e-07,
+      "loss": 1.6594,
+      "step": 3135
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.444473311507582e-07,
+      "loss": 1.6225,
+      "step": 3136
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.439060053787617e-07,
+      "loss": 1.6542,
+      "step": 3137
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.433651963898813e-07,
+      "loss": 1.6712,
+      "step": 3138
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.428249045537546e-07,
+      "loss": 1.6619,
+      "step": 3139
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.422851302396655e-07,
+      "loss": 1.6642,
+      "step": 3140
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.417458738165451e-07,
+      "loss": 1.669,
+      "step": 3141
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4120713565297003e-07,
+      "loss": 1.653,
+      "step": 3142
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.406689161171628e-07,
+      "loss": 1.6816,
+      "step": 3143
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4013121557699157e-07,
+      "loss": 1.6295,
+      "step": 3144
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.395940343999691e-07,
+      "loss": 1.6574,
+      "step": 3145
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.390573729532539e-07,
+      "loss": 1.6335,
+      "step": 3146
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.385212316036491e-07,
+      "loss": 1.612,
+      "step": 3147
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.3798561071760238e-07,
+      "loss": 1.6576,
+      "step": 3148
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.3745051066120535e-07,
+      "loss": 1.6361,
+      "step": 3149
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.3691593180019366e-07,
+      "loss": 1.6703,
+      "step": 3150
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.3638187449994717e-07,
+      "loss": 1.6357,
+      "step": 3151
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.3584833912548885e-07,
+      "loss": 1.6527,
+      "step": 3152
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.353153260414853e-07,
+      "loss": 1.6422,
+      "step": 3153
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.347828356122461e-07,
+      "loss": 1.6479,
+      "step": 3154
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.342508682017229e-07,
+      "loss": 1.6585,
+      "step": 3155
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.3371942417351076e-07,
+      "loss": 1.6328,
+      "step": 3156
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.3318850389084656e-07,
+      "loss": 1.6249,
+      "step": 3157
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.3265810771660966e-07,
+      "loss": 1.6262,
+      "step": 3158
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.3212823601332055e-07,
+      "loss": 1.6451,
+      "step": 3159
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.3159888914314119e-07,
+      "loss": 1.6797,
+      "step": 3160
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.310700674678755e-07,
+      "loss": 1.6573,
+      "step": 3161
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.3054177134896812e-07,
+      "loss": 1.6725,
+      "step": 3162
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.3001400114750458e-07,
+      "loss": 1.679,
+      "step": 3163
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2948675722421085e-07,
+      "loss": 1.6512,
+      "step": 3164
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.289600399394529e-07,
+      "loss": 1.6423,
+      "step": 3165
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2843384965323708e-07,
+      "loss": 1.6531,
+      "step": 3166
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2790818672520996e-07,
+      "loss": 1.6417,
+      "step": 3167
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2738305151465642e-07,
+      "loss": 1.6628,
+      "step": 3168
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2685844438050216e-07,
+      "loss": 1.6677,
+      "step": 3169
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.263343656813107e-07,
+      "loss": 1.6662,
+      "step": 3170
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2581081577528494e-07,
+      "loss": 1.6626,
+      "step": 3171
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.252877950202665e-07,
+      "loss": 1.6649,
+      "step": 3172
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2476530377373503e-07,
+      "loss": 1.6252,
+      "step": 3173
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2424334239280872e-07,
+      "loss": 1.6318,
+      "step": 3174
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2372191123424256e-07,
+      "loss": 1.6562,
+      "step": 3175
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2320101065443054e-07,
+      "loss": 1.6418,
+      "step": 3176
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2268064100940276e-07,
+      "loss": 1.6478,
+      "step": 3177
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2216080265482706e-07,
+      "loss": 1.6538,
+      "step": 3178
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2164149594600823e-07,
+      "loss": 1.6512,
+      "step": 3179
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2112272123788768e-07,
+      "loss": 1.6005,
+      "step": 3180
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.2060447888504251e-07,
+      "loss": 1.6488,
+      "step": 3181
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.2008676924168678e-07,
+      "loss": 1.668,
+      "step": 3182
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.195695926616702e-07,
+      "loss": 1.6474,
+      "step": 3183
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.190529494984782e-07,
+      "loss": 1.6129,
+      "step": 3184
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.18536840105231e-07,
+      "loss": 1.6507,
+      "step": 3185
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.180212648346852e-07,
+      "loss": 1.6334,
+      "step": 3186
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.1750622403923103e-07,
+      "loss": 1.6469,
+      "step": 3187
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.1699171807089411e-07,
+      "loss": 1.6812,
+      "step": 3188
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.164777472813347e-07,
+      "loss": 1.6357,
+      "step": 3189
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.1596431202184705e-07,
+      "loss": 1.6452,
+      "step": 3190
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.1545141264335887e-07,
+      "loss": 1.651,
+      "step": 3191
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.1493904949643225e-07,
+      "loss": 1.6599,
+      "step": 3192
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.144272229312627e-07,
+      "loss": 1.6435,
+      "step": 3193
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.1391593329767897e-07,
+      "loss": 1.6566,
+      "step": 3194
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.134051809451426e-07,
+      "loss": 1.6438,
+      "step": 3195
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.128949662227475e-07,
+      "loss": 1.6616,
+      "step": 3196
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.123852894792211e-07,
+      "loss": 1.6498,
+      "step": 3197
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.1187615106292256e-07,
+      "loss": 1.6296,
+      "step": 3198
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.1136755132184325e-07,
+      "loss": 1.6475,
+      "step": 3199
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.1085949060360653e-07,
+      "loss": 1.6279,
+      "step": 3200
+    },
+    {
+      "epoch": 2.25,
+      "eval_loss": 1.6541236639022827,
+      "eval_runtime": 43.0871,
+      "eval_samples_per_second": 46.418,
+      "eval_steps_per_second": 0.116,
+      "step": 3200
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.103519692554665e-07,
+      "loss": 1.6427,
+      "step": 3201
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.0984498762430957e-07,
+      "loss": 1.6553,
+      "step": 3202
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.093385460566529e-07,
+      "loss": 1.6625,
+      "step": 3203
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.088326448986447e-07,
+      "loss": 1.6595,
+      "step": 3204
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.0832728449606364e-07,
+      "loss": 1.6676,
+      "step": 3205
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.0782246519431813e-07,
+      "loss": 1.6489,
+      "step": 3206
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.0731818733844797e-07,
+      "loss": 1.6271,
+      "step": 3207
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.0681445127312213e-07,
+      "loss": 1.6191,
+      "step": 3208
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.0631125734263966e-07,
+      "loss": 1.6339,
+      "step": 3209
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.0580860589092896e-07,
+      "loss": 1.6441,
+      "step": 3210
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.0530649726154715e-07,
+      "loss": 1.6356,
+      "step": 3211
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.048049317976809e-07,
+      "loss": 1.6505,
+      "step": 3212
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.0430390984214585e-07,
+      "loss": 1.647,
+      "step": 3213
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.0380343173738513e-07,
+      "loss": 1.6001,
+      "step": 3214
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.033034978254714e-07,
+      "loss": 1.6224,
+      "step": 3215
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.0280410844810424e-07,
+      "loss": 1.6575,
+      "step": 3216
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.0230526394661185e-07,
+      "loss": 1.6765,
+      "step": 3217
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.018069646619498e-07,
+      "loss": 1.6561,
+      "step": 3218
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.0130921093470087e-07,
+      "loss": 1.642,
+      "step": 3219
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.0081200310507528e-07,
+      "loss": 1.6503,
+      "step": 3220
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.0031534151290942e-07,
+      "loss": 1.6378,
+      "step": 3221
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9981922649766714e-07,
+      "loss": 1.6354,
+      "step": 3222
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9932365839843844e-07,
+      "loss": 1.6463,
+      "step": 3223
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9882863755393908e-07,
+      "loss": 1.6502,
+      "step": 3224
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.983341643025117e-07,
+      "loss": 1.6418,
+      "step": 3225
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9784023898212342e-07,
+      "loss": 1.6365,
+      "step": 3226
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9734686193036786e-07,
+      "loss": 1.6347,
+      "step": 3227
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9685403348446373e-07,
+      "loss": 1.6361,
+      "step": 3228
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9636175398125432e-07,
+      "loss": 1.6548,
+      "step": 3229
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9587002375720862e-07,
+      "loss": 1.6491,
+      "step": 3230
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9537884314841879e-07,
+      "loss": 1.6601,
+      "step": 3231
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9488821249060293e-07,
+      "loss": 1.6685,
+      "step": 3232
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9439813211910172e-07,
+      "loss": 1.6354,
+      "step": 3233
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9390860236888095e-07,
+      "loss": 1.6447,
+      "step": 3234
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9341962357452967e-07,
+      "loss": 1.6541,
+      "step": 3235
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9293119607025987e-07,
+      "loss": 1.6347,
+      "step": 3236
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9244332018990727e-07,
+      "loss": 1.6291,
+      "step": 3237
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9195599626693048e-07,
+      "loss": 1.6521,
+      "step": 3238
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9146922463441095e-07,
+      "loss": 1.6605,
+      "step": 3239
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9098300562505264e-07,
+      "loss": 1.6427,
+      "step": 3240
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9049733957118142e-07,
+      "loss": 1.6349,
+      "step": 3241
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9001222680474517e-07,
+      "loss": 1.6708,
+      "step": 3242
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.895276676573141e-07,
+      "loss": 1.6441,
+      "step": 3243
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.8904366246007997e-07,
+      "loss": 1.6516,
+      "step": 3244
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.8856021154385594e-07,
+      "loss": 1.649,
+      "step": 3245
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.8807731523907555e-07,
+      "loss": 1.6617,
+      "step": 3246
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.8759497387579426e-07,
+      "loss": 1.6373,
+      "step": 3247
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.8711318778368789e-07,
+      "loss": 1.6643,
+      "step": 3248
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.8663195729205262e-07,
+      "loss": 1.6413,
+      "step": 3249
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.8615128272980507e-07,
+      "loss": 1.6476,
+      "step": 3250
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.856711644254817e-07,
+      "loss": 1.6552,
+      "step": 3251
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.8519160270723854e-07,
+      "loss": 1.6433,
+      "step": 3252
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.8471259790285177e-07,
+      "loss": 1.6452,
+      "step": 3253
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.8423415033971656e-07,
+      "loss": 1.6408,
+      "step": 3254
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.837562603448477e-07,
+      "loss": 1.6687,
+      "step": 3255
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.832789282448779e-07,
+      "loss": 1.6414,
+      "step": 3256
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.828021543660594e-07,
+      "loss": 1.671,
+      "step": 3257
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.8232593903426262e-07,
+      "loss": 1.6541,
+      "step": 3258
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.8185028257497636e-07,
+      "loss": 1.6541,
+      "step": 3259
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.8137518531330764e-07,
+      "loss": 1.6357,
+      "step": 3260
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.809006475739805e-07,
+      "loss": 1.6513,
+      "step": 3261
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.804266696813368e-07,
+      "loss": 1.6161,
+      "step": 3262
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.7995325195933631e-07,
+      "loss": 1.6834,
+      "step": 3263
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.794803947315555e-07,
+      "loss": 1.6291,
+      "step": 3264
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.7900809832118814e-07,
+      "loss": 1.6399,
+      "step": 3265
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.785363630510438e-07,
+      "loss": 1.6558,
+      "step": 3266
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7806518924354929e-07,
+      "loss": 1.6404,
+      "step": 3267
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7759457722074766e-07,
+      "loss": 1.644,
+      "step": 3268
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7712452730429793e-07,
+      "loss": 1.6635,
+      "step": 3269
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7665503981547425e-07,
+      "loss": 1.6933,
+      "step": 3270
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7618611507516746e-07,
+      "loss": 1.6658,
+      "step": 3271
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7571775340388272e-07,
+      "loss": 1.6532,
+      "step": 3272
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.752499551217409e-07,
+      "loss": 1.6099,
+      "step": 3273
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7478272054847787e-07,
+      "loss": 1.677,
+      "step": 3274
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7431605000344428e-07,
+      "loss": 1.6228,
+      "step": 3275
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7384994380560448e-07,
+      "loss": 1.658,
+      "step": 3276
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.73384402273538e-07,
+      "loss": 1.6329,
+      "step": 3277
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7291942572543805e-07,
+      "loss": 1.6523,
+      "step": 3278
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7245501447911192e-07,
+      "loss": 1.6449,
+      "step": 3279
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7199116885197996e-07,
+      "loss": 1.6336,
+      "step": 3280
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.7152788916107662e-07,
+      "loss": 1.6579,
+      "step": 3281
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.710651757230488e-07,
+      "loss": 1.6418,
+      "step": 3282
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.706030288541569e-07,
+      "loss": 1.6542,
+      "step": 3283
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.7014144887027405e-07,
+      "loss": 1.6557,
+      "step": 3284
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.696804360868861e-07,
+      "loss": 1.6333,
+      "step": 3285
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.6921999081909033e-07,
+      "loss": 1.6467,
+      "step": 3286
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.6876011338159713e-07,
+      "loss": 1.6507,
+      "step": 3287
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.683008040887285e-07,
+      "loss": 1.6492,
+      "step": 3288
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.6784206325441764e-07,
+      "loss": 1.643,
+      "step": 3289
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.6738389119220963e-07,
+      "loss": 1.6914,
+      "step": 3290
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.6692628821526123e-07,
+      "loss": 1.6256,
+      "step": 3291
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.664692546363392e-07,
+      "loss": 1.6427,
+      "step": 3292
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.660127907678218e-07,
+      "loss": 1.6475,
+      "step": 3293
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.6555689692169795e-07,
+      "loss": 1.6624,
+      "step": 3294
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.6510157340956698e-07,
+      "loss": 1.6205,
+      "step": 3295
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.6464682054263767e-07,
+      "loss": 1.6377,
+      "step": 3296
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.6419263863172993e-07,
+      "loss": 1.6342,
+      "step": 3297
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.6373902798727225e-07,
+      "loss": 1.6732,
+      "step": 3298
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.6328598891930334e-07,
+      "loss": 1.6507,
+      "step": 3299
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.6283352173747146e-07,
+      "loss": 1.6569,
+      "step": 3300
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.623816267510335e-07,
+      "loss": 1.5896,
+      "step": 3301
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.6193030426885523e-07,
+      "loss": 1.6442,
+      "step": 3302
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.6147955459941153e-07,
+      "loss": 1.6655,
+      "step": 3303
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.6102937805078542e-07,
+      "loss": 1.6647,
+      "step": 3304
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.6057977493066866e-07,
+      "loss": 1.6615,
+      "step": 3305
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.601307455463602e-07,
+      "loss": 1.6546,
+      "step": 3306
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5968229020476797e-07,
+      "loss": 1.6377,
+      "step": 3307
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5923440921240639e-07,
+      "loss": 1.6423,
+      "step": 3308
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.587871028753981e-07,
+      "loss": 1.6647,
+      "step": 3309
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5834037149947288e-07,
+      "loss": 1.6755,
+      "step": 3310
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.578942153899676e-07,
+      "loss": 1.6787,
+      "step": 3311
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5744863485182535e-07,
+      "loss": 1.6584,
+      "step": 3312
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5700363018959628e-07,
+      "loss": 1.6381,
+      "step": 3313
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5655920170743708e-07,
+      "loss": 1.6355,
+      "step": 3314
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5611534970911056e-07,
+      "loss": 1.6378,
+      "step": 3315
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5567207449798515e-07,
+      "loss": 1.6213,
+      "step": 3316
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5522937637703504e-07,
+      "loss": 1.6472,
+      "step": 3317
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5478725564884044e-07,
+      "loss": 1.6563,
+      "step": 3318
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5434571261558682e-07,
+      "loss": 1.6253,
+      "step": 3319
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5390474757906448e-07,
+      "loss": 1.6825,
+      "step": 3320
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5346436084066928e-07,
+      "loss": 1.6478,
+      "step": 3321
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.530245527014008e-07,
+      "loss": 1.643,
+      "step": 3322
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5258532346186403e-07,
+      "loss": 1.6515,
+      "step": 3323
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5214667342226816e-07,
+      "loss": 1.6396,
+      "step": 3324
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5170860288242636e-07,
+      "loss": 1.6362,
+      "step": 3325
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5127111214175558e-07,
+      "loss": 1.628,
+      "step": 3326
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.508342014992764e-07,
+      "loss": 1.6543,
+      "step": 3327
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5039787125361326e-07,
+      "loss": 1.6748,
+      "step": 3328
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.4996212170299393e-07,
+      "loss": 1.6742,
+      "step": 3329
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.495269531452491e-07,
+      "loss": 1.6267,
+      "step": 3330
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.4909236587781248e-07,
+      "loss": 1.6721,
+      "step": 3331
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.4865836019771993e-07,
+      "loss": 1.6385,
+      "step": 3332
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.482249364016106e-07,
+      "loss": 1.6783,
+      "step": 3333
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.4779209478572542e-07,
+      "loss": 1.6224,
+      "step": 3334
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.473598356459078e-07,
+      "loss": 1.6458,
+      "step": 3335
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.469281592776027e-07,
+      "loss": 1.6528,
+      "step": 3336
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.4649706597585643e-07,
+      "loss": 1.6437,
+      "step": 3337
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.4606655603531758e-07,
+      "loss": 1.649,
+      "step": 3338
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.456366297502355e-07,
+      "loss": 1.6322,
+      "step": 3339
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.4520728741446086e-07,
+      "loss": 1.6516,
+      "step": 3340
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.4477852932144542e-07,
+      "loss": 1.639,
+      "step": 3341
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.443503557642405e-07,
+      "loss": 1.6328,
+      "step": 3342
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.4392276703549922e-07,
+      "loss": 1.6484,
+      "step": 3343
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.434957634274746e-07,
+      "loss": 1.6393,
+      "step": 3344
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.4306934523201898e-07,
+      "loss": 1.6166,
+      "step": 3345
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.4264351274058583e-07,
+      "loss": 1.6778,
+      "step": 3346
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.4221826624422707e-07,
+      "loss": 1.6361,
+      "step": 3347
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.4179360603359503e-07,
+      "loss": 1.6641,
+      "step": 3348
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.413695323989408e-07,
+      "loss": 1.6496,
+      "step": 3349
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.409460456301147e-07,
+      "loss": 1.6487,
+      "step": 3350
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.4052314601656634e-07,
+      "loss": 1.6466,
+      "step": 3351
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.4010083384734306e-07,
+      "loss": 1.6344,
+      "step": 3352
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.3967910941109163e-07,
+      "loss": 1.6438,
+      "step": 3353
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.3925797299605645e-07,
+      "loss": 1.6512,
+      "step": 3354
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.3883742489008032e-07,
+      "loss": 1.6538,
+      "step": 3355
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.384174653806044e-07,
+      "loss": 1.6205,
+      "step": 3356
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.379980947546663e-07,
+      "loss": 1.6393,
+      "step": 3357
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.3757931329890217e-07,
+      "loss": 1.6534,
+      "step": 3358
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.3716112129954538e-07,
+      "loss": 1.6704,
+      "step": 3359
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.3674351904242608e-07,
+      "loss": 1.654,
+      "step": 3360
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.3632650681297164e-07,
+      "loss": 1.6594,
+      "step": 3361
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.359100848962057e-07,
+      "loss": 1.621,
+      "step": 3362
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.3549425357674905e-07,
+      "loss": 1.6486,
+      "step": 3363
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.350790131388181e-07,
+      "loss": 1.6438,
+      "step": 3364
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.3466436386622583e-07,
+      "loss": 1.6304,
+      "step": 3365
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.3425030604238153e-07,
+      "loss": 1.6255,
+      "step": 3366
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.3383683995028926e-07,
+      "loss": 1.6479,
+      "step": 3367
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.3342396587254957e-07,
+      "loss": 1.6579,
+      "step": 3368
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.3301168409135776e-07,
+      "loss": 1.6356,
+      "step": 3369
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.325999948885047e-07,
+      "loss": 1.6226,
+      "step": 3370
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.321888985453763e-07,
+      "loss": 1.6347,
+      "step": 3371
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.3177839534295277e-07,
+      "loss": 1.6444,
+      "step": 3372
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.313684855618089e-07,
+      "loss": 1.652,
+      "step": 3373
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.3095916948211448e-07,
+      "loss": 1.6422,
+      "step": 3374
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.3055044738363307e-07,
+      "loss": 1.6169,
+      "step": 3375
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.3014231954572286e-07,
+      "loss": 1.6445,
+      "step": 3376
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.2973478624733458e-07,
+      "loss": 1.6248,
+      "step": 3377
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.2932784776701388e-07,
+      "loss": 1.6232,
+      "step": 3378
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.2892150438289928e-07,
+      "loss": 1.6585,
+      "step": 3379
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.285157563727226e-07,
+      "loss": 1.6613,
+      "step": 3380
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.2811060401380912e-07,
+      "loss": 1.6406,
+      "step": 3381
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.2770604758307634e-07,
+      "loss": 1.6366,
+      "step": 3382
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.273020873570345e-07,
+      "loss": 1.6287,
+      "step": 3383
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.26898723611787e-07,
+      "loss": 1.6039,
+      "step": 3384
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.2649595662302905e-07,
+      "loss": 1.6393,
+      "step": 3385
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.260937866660483e-07,
+      "loss": 1.6581,
+      "step": 3386
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.2569221401572373e-07,
+      "loss": 1.6629,
+      "step": 3387
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.252912389465266e-07,
+      "loss": 1.6478,
+      "step": 3388
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.2489086173251973e-07,
+      "loss": 1.654,
+      "step": 3389
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.244910826473572e-07,
+      "loss": 1.6244,
+      "step": 3390
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.24091901964284e-07,
+      "loss": 1.6294,
+      "step": 3391
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.2369331995613663e-07,
+      "loss": 1.6511,
+      "step": 3392
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.2329533689534177e-07,
+      "loss": 1.633,
+      "step": 3393
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.2289795305391725e-07,
+      "loss": 1.6279,
+      "step": 3394
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.225011687034714e-07,
+      "loss": 1.6222,
+      "step": 3395
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.2210498411520253e-07,
+      "loss": 1.6502,
+      "step": 3396
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.217093995598988e-07,
+      "loss": 1.6794,
+      "step": 3397
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.213144153079386e-07,
+      "loss": 1.6348,
+      "step": 3398
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.2092003162929009e-07,
+      "loss": 1.6239,
+      "step": 3399
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.2052624879351103e-07,
+      "loss": 1.6353,
+      "step": 3400
+    },
+    {
+      "epoch": 2.39,
+      "eval_loss": 1.6535223722457886,
+      "eval_runtime": 44.0421,
+      "eval_samples_per_second": 45.411,
+      "eval_steps_per_second": 0.114,
+      "step": 3400
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.201330670697477e-07,
+      "loss": 1.6463,
+      "step": 3401
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1974048672673654e-07,
+      "loss": 1.6482,
+      "step": 3402
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1934850803280215e-07,
+      "loss": 1.6558,
+      "step": 3403
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1895713125585849e-07,
+      "loss": 1.6658,
+      "step": 3404
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1856635666340786e-07,
+      "loss": 1.6324,
+      "step": 3405
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1817618452254108e-07,
+      "loss": 1.6149,
+      "step": 3406
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1778661509993726e-07,
+      "loss": 1.6286,
+      "step": 3407
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1739764866186308e-07,
+      "loss": 1.6336,
+      "step": 3408
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1700928547417377e-07,
+      "loss": 1.6345,
+      "step": 3409
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1662152580231144e-07,
+      "loss": 1.6564,
+      "step": 3410
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1623436991130653e-07,
+      "loss": 1.636,
+      "step": 3411
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1584781806577659e-07,
+      "loss": 1.6324,
+      "step": 3412
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1546187052992584e-07,
+      "loss": 1.6456,
+      "step": 3413
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1507652756754571e-07,
+      "loss": 1.6585,
+      "step": 3414
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1469178944201474e-07,
+      "loss": 1.6224,
+      "step": 3415
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1430765641629769e-07,
+      "loss": 1.635,
+      "step": 3416
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1392412875294622e-07,
+      "loss": 1.6555,
+      "step": 3417
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1354120671409739e-07,
+      "loss": 1.6531,
+      "step": 3418
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1315889056147487e-07,
+      "loss": 1.6232,
+      "step": 3419
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1277718055638819e-07,
+      "loss": 1.6305,
+      "step": 3420
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1239607695973263e-07,
+      "loss": 1.6362,
+      "step": 3421
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1201558003198907e-07,
+      "loss": 1.621,
+      "step": 3422
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1163569003322316e-07,
+      "loss": 1.6318,
+      "step": 3423
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1125640722308626e-07,
+      "loss": 1.6636,
+      "step": 3424
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1087773186081473e-07,
+      "loss": 1.6342,
+      "step": 3425
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.104996642052296e-07,
+      "loss": 1.6936,
+      "step": 3426
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1012220451473664e-07,
+      "loss": 1.6445,
+      "step": 3427
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.097453530473258e-07,
+      "loss": 1.6516,
+      "step": 3428
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0936911006057148e-07,
+      "loss": 1.6488,
+      "step": 3429
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.089934758116322e-07,
+      "loss": 1.612,
+      "step": 3430
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.086184505572505e-07,
+      "loss": 1.6385,
+      "step": 3431
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0824403455375286e-07,
+      "loss": 1.6373,
+      "step": 3432
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.078702280570486e-07,
+      "loss": 1.6247,
+      "step": 3433
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0749703132263122e-07,
+      "loss": 1.6529,
+      "step": 3434
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0712444460557712e-07,
+      "loss": 1.6253,
+      "step": 3435
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0675246816054584e-07,
+      "loss": 1.6625,
+      "step": 3436
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0638110224177999e-07,
+      "loss": 1.629,
+      "step": 3437
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.060103471031043e-07,
+      "loss": 1.6484,
+      "step": 3438
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0564020299792654e-07,
+      "loss": 1.6653,
+      "step": 3439
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0527067017923652e-07,
+      "loss": 1.6632,
+      "step": 3440
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.049017488996068e-07,
+      "loss": 1.6664,
+      "step": 3441
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0453343941119153e-07,
+      "loss": 1.6103,
+      "step": 3442
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0416574196572648e-07,
+      "loss": 1.6514,
+      "step": 3443
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0379865681452971e-07,
+      "loss": 1.6335,
+      "step": 3444
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0343218420850019e-07,
+      "loss": 1.642,
+      "step": 3445
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0306632439811891e-07,
+      "loss": 1.6706,
+      "step": 3446
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0270107763344704e-07,
+      "loss": 1.6366,
+      "step": 3447
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.023364441641279e-07,
+      "loss": 1.6504,
+      "step": 3448
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0197242423938445e-07,
+      "loss": 1.6438,
+      "step": 3449
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0160901810802114e-07,
+      "loss": 1.628,
+      "step": 3450
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0124622601842269e-07,
+      "loss": 1.6575,
+      "step": 3451
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0088404821855412e-07,
+      "loss": 1.635,
+      "step": 3452
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0052248495596027e-07,
+      "loss": 1.6607,
+      "step": 3453
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0016153647776637e-07,
+      "loss": 1.6254,
+      "step": 3454
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.98012030306774e-08,
+      "loss": 1.6686,
+      "step": 3455
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.944148486097792e-08,
+      "loss": 1.6545,
+      "step": 3456
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.908238221453158e-08,
+      "loss": 1.6562,
+      "step": 3457
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.872389533678205e-08,
+      "loss": 1.6738,
+      "step": 3458
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.836602447275144e-08,
+      "loss": 1.6582,
+      "step": 3459
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.800876986704109e-08,
+      "loss": 1.6815,
+      "step": 3460
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.765213176383125e-08,
+      "loss": 1.664,
+      "step": 3461
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.729611040688101e-08,
+      "loss": 1.6325,
+      "step": 3462
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.694070603952698e-08,
+      "loss": 1.6598,
+      "step": 3463
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.658591890468514e-08,
+      "loss": 1.6406,
+      "step": 3464
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.623174924484922e-08,
+      "loss": 1.6543,
+      "step": 3465
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.587819730209046e-08,
+      "loss": 1.6473,
+      "step": 3466
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.552526331805855e-08,
+      "loss": 1.6501,
+      "step": 3467
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.517294753398064e-08,
+      "loss": 1.6261,
+      "step": 3468
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.482125019066078e-08,
+      "loss": 1.6493,
+      "step": 3469
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.447017152848124e-08,
+      "loss": 1.643,
+      "step": 3470
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.411971178740075e-08,
+      "loss": 1.6623,
+      "step": 3471
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.376987120695545e-08,
+      "loss": 1.6367,
+      "step": 3472
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.342065002625788e-08,
+      "loss": 1.6688,
+      "step": 3473
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.307204848399753e-08,
+      "loss": 1.6553,
+      "step": 3474
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.272406681844014e-08,
+      "loss": 1.653,
+      "step": 3475
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.237670526742791e-08,
+      "loss": 1.6297,
+      "step": 3476
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.202996406837937e-08,
+      "loss": 1.6294,
+      "step": 3477
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.168384345828883e-08,
+      "loss": 1.6191,
+      "step": 3478
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.133834367372618e-08,
+      "loss": 1.625,
+      "step": 3479
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.099346495083749e-08,
+      "loss": 1.6032,
+      "step": 3480
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.06492075253441e-08,
+      "loss": 1.6471,
+      "step": 3481
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.030557163254282e-08,
+      "loss": 1.6811,
+      "step": 3482
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 8.996255750730531e-08,
+      "loss": 1.6499,
+      "step": 3483
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 8.96201653840788e-08,
+      "loss": 1.6792,
+      "step": 3484
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 8.927839549688465e-08,
+      "loss": 1.6891,
+      "step": 3485
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 8.893724807931968e-08,
+      "loss": 1.6266,
+      "step": 3486
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 8.85967233645547e-08,
+      "loss": 1.651,
+      "step": 3487
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 8.825682158533553e-08,
+      "loss": 1.6369,
+      "step": 3488
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 8.791754297398147e-08,
+      "loss": 1.6405,
+      "step": 3489
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 8.75788877623862e-08,
+      "loss": 1.6361,
+      "step": 3490
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 8.72408561820177e-08,
+      "loss": 1.6507,
+      "step": 3491
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 8.690344846391729e-08,
+      "loss": 1.653,
+      "step": 3492
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 8.65666648386999e-08,
+      "loss": 1.6586,
+      "step": 3493
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.623050553655365e-08,
+      "loss": 1.6399,
+      "step": 3494
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.589497078724061e-08,
+      "loss": 1.6589,
+      "step": 3495
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.556006082009559e-08,
+      "loss": 1.6757,
+      "step": 3496
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.522577586402624e-08,
+      "loss": 1.6542,
+      "step": 3497
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.489211614751357e-08,
+      "loss": 1.6406,
+      "step": 3498
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.455908189861038e-08,
+      "loss": 1.6578,
+      "step": 3499
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.422667334494249e-08,
+      "loss": 1.6613,
+      "step": 3500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.389489071370825e-08,
+      "loss": 1.6307,
+      "step": 3501
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.356373423167806e-08,
+      "loss": 1.6692,
+      "step": 3502
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.323320412519397e-08,
+      "loss": 1.6523,
+      "step": 3503
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.290330062017014e-08,
+      "loss": 1.6664,
+      "step": 3504
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.257402394209257e-08,
+      "loss": 1.6737,
+      "step": 3505
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.224537431601886e-08,
+      "loss": 1.6812,
+      "step": 3506
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.191735196657789e-08,
+      "loss": 1.675,
+      "step": 3507
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.158995711797e-08,
+      "loss": 1.6814,
+      "step": 3508
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.126318999396619e-08,
+      "loss": 1.6693,
+      "step": 3509
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.093705081790891e-08,
+      "loss": 1.6375,
+      "step": 3510
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.061153981271107e-08,
+      "loss": 1.6417,
+      "step": 3511
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.028665720085659e-08,
+      "loss": 1.6576,
+      "step": 3512
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.996240320439962e-08,
+      "loss": 1.6513,
+      "step": 3513
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.963877804496444e-08,
+      "loss": 1.6515,
+      "step": 3514
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.9315781943746e-08,
+      "loss": 1.6488,
+      "step": 3515
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.899341512150892e-08,
+      "loss": 1.6245,
+      "step": 3516
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.867167779858819e-08,
+      "loss": 1.6598,
+      "step": 3517
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.835057019488823e-08,
+      "loss": 1.6517,
+      "step": 3518
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.803009252988268e-08,
+      "loss": 1.6363,
+      "step": 3519
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.771024502261525e-08,
+      "loss": 1.6389,
+      "step": 3520
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.739102789169882e-08,
+      "loss": 1.656,
+      "step": 3521
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.707244135531498e-08,
+      "loss": 1.6578,
+      "step": 3522
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.6754485631215e-08,
+      "loss": 1.6324,
+      "step": 3523
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.643716093671826e-08,
+      "loss": 1.6257,
+      "step": 3524
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.612046748871326e-08,
+      "loss": 1.6922,
+      "step": 3525
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.580440550365707e-08,
+      "loss": 1.6657,
+      "step": 3526
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.548897519757513e-08,
+      "loss": 1.6582,
+      "step": 3527
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.51741767860612e-08,
+      "loss": 1.6315,
+      "step": 3528
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.486001048427671e-08,
+      "loss": 1.6567,
+      "step": 3529
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.454647650695157e-08,
+      "loss": 1.6592,
+      "step": 3530
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.423357506838312e-08,
+      "loss": 1.6588,
+      "step": 3531
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.392130638243665e-08,
+      "loss": 1.6337,
+      "step": 3532
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.360967066254508e-08,
+      "loss": 1.6552,
+      "step": 3533
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.329866812170805e-08,
+      "loss": 1.6671,
+      "step": 3534
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.298829897249303e-08,
+      "loss": 1.6374,
+      "step": 3535
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.26785634270346e-08,
+      "loss": 1.6526,
+      "step": 3536
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.236946169703384e-08,
+      "loss": 1.649,
+      "step": 3537
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.20609939937592e-08,
+      "loss": 1.6621,
+      "step": 3538
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.175316052804525e-08,
+      "loss": 1.6586,
+      "step": 3539
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.144596151029303e-08,
+      "loss": 1.6326,
+      "step": 3540
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.113939715047034e-08,
+      "loss": 1.6462,
+      "step": 3541
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.083346765811115e-08,
+      "loss": 1.656,
+      "step": 3542
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.052817324231541e-08,
+      "loss": 1.6683,
+      "step": 3543
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.022351411174865e-08,
+      "loss": 1.6493,
+      "step": 3544
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.991949047464285e-08,
+      "loss": 1.6218,
+      "step": 3545
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.961610253879502e-08,
+      "loss": 1.6091,
+      "step": 3546
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.931335051156828e-08,
+      "loss": 1.6319,
+      "step": 3547
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.901123459989066e-08,
+      "loss": 1.6481,
+      "step": 3548
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.870975501025555e-08,
+      "loss": 1.6362,
+      "step": 3549
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.840891194872111e-08,
+      "loss": 1.6196,
+      "step": 3550
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.810870562091109e-08,
+      "loss": 1.656,
+      "step": 3551
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.780913623201345e-08,
+      "loss": 1.6475,
+      "step": 3552
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.751020398678143e-08,
+      "loss": 1.6339,
+      "step": 3553
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.721190908953189e-08,
+      "loss": 1.6484,
+      "step": 3554
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.691425174414678e-08,
+      "loss": 1.6748,
+      "step": 3555
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.661723215407222e-08,
+      "loss": 1.6607,
+      "step": 3556
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.632085052231817e-08,
+      "loss": 1.6578,
+      "step": 3557
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.602510705145892e-08,
+      "loss": 1.6336,
+      "step": 3558
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.573000194363221e-08,
+      "loss": 1.6513,
+      "step": 3559
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.543553540053926e-08,
+      "loss": 1.6321,
+      "step": 3560
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.514170762344539e-08,
+      "loss": 1.6112,
+      "step": 3561
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.484851881317932e-08,
+      "loss": 1.628,
+      "step": 3562
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.455596917013273e-08,
+      "loss": 1.6668,
+      "step": 3563
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.426405889426046e-08,
+      "loss": 1.6748,
+      "step": 3564
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.397278818508034e-08,
+      "loss": 1.6284,
+      "step": 3565
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.368215724167336e-08,
+      "loss": 1.6629,
+      "step": 3566
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.339216626268318e-08,
+      "loss": 1.6439,
+      "step": 3567
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.310281544631546e-08,
+      "loss": 1.6508,
+      "step": 3568
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.281410499033912e-08,
+      "loss": 1.6525,
+      "step": 3569
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.252603509208465e-08,
+      "loss": 1.6286,
+      "step": 3570
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.223860594844544e-08,
+      "loss": 1.6332,
+      "step": 3571
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.195181775587654e-08,
+      "loss": 1.6542,
+      "step": 3572
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.166567071039496e-08,
+      "loss": 1.6745,
+      "step": 3573
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.138016500757948e-08,
+      "loss": 1.6609,
+      "step": 3574
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.109530084257042e-08,
+      "loss": 1.6674,
+      "step": 3575
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.081107841007004e-08,
+      "loss": 1.6327,
+      "step": 3576
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.052749790434153e-08,
+      "loss": 1.6401,
+      "step": 3577
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.024455951920949e-08,
+      "loss": 1.6222,
+      "step": 3578
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.996226344805978e-08,
+      "loss": 1.6348,
+      "step": 3579
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.968060988383883e-08,
+      "loss": 1.6394,
+      "step": 3580
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.939959901905423e-08,
+      "loss": 1.6446,
+      "step": 3581
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.911923104577454e-08,
+      "loss": 1.6488,
+      "step": 3582
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.883950615562849e-08,
+      "loss": 1.6468,
+      "step": 3583
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.8560424539805255e-08,
+      "loss": 1.654,
+      "step": 3584
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.828198638905457e-08,
+      "loss": 1.6537,
+      "step": 3585
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.800419189368655e-08,
+      "loss": 1.6617,
+      "step": 3586
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.772704124357064e-08,
+      "loss": 1.6131,
+      "step": 3587
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.745053462813698e-08,
+      "loss": 1.6652,
+      "step": 3588
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.7174672236375286e-08,
+      "loss": 1.6597,
+      "step": 3589
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.6899454256834734e-08,
+      "loss": 1.6216,
+      "step": 3590
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.662488087762429e-08,
+      "loss": 1.6488,
+      "step": 3591
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.635095228641229e-08,
+      "loss": 1.6461,
+      "step": 3592
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.6077668670426407e-08,
+      "loss": 1.6692,
+      "step": 3593
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.580503021645322e-08,
+      "loss": 1.6228,
+      "step": 3594
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.553303711083879e-08,
+      "loss": 1.646,
+      "step": 3595
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.526168953948751e-08,
+      "loss": 1.6554,
+      "step": 3596
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.4990987687863146e-08,
+      "loss": 1.6367,
+      "step": 3597
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.4720931740987796e-08,
+      "loss": 1.6398,
+      "step": 3598
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.445152188344226e-08,
+      "loss": 1.6398,
+      "step": 3599
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.4182758299365364e-08,
+      "loss": 1.6181,
+      "step": 3600
+    },
+    {
+      "epoch": 2.53,
+      "eval_loss": 1.6532448530197144,
+      "eval_runtime": 45.5622,
+      "eval_samples_per_second": 43.896,
+      "eval_steps_per_second": 0.11,
+      "step": 3600
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.39146411724547e-08,
+      "loss": 1.6178,
+      "step": 3601
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.3647170685965693e-08,
+      "loss": 1.6598,
+      "step": 3602
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.3380347022712104e-08,
+      "loss": 1.664,
+      "step": 3603
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.311417036506516e-08,
+      "loss": 1.6574,
+      "step": 3604
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.2848640894954223e-08,
+      "loss": 1.6414,
+      "step": 3605
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.2583758793866005e-08,
+      "loss": 1.6633,
+      "step": 3606
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.2319524242845135e-08,
+      "loss": 1.637,
+      "step": 3607
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.2055937422493254e-08,
+      "loss": 1.6361,
+      "step": 3608
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.179299851296981e-08,
+      "loss": 1.594,
+      "step": 3609
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.1530707693990594e-08,
+      "loss": 1.6419,
+      "step": 3610
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.1269065144829205e-08,
+      "loss": 1.644,
+      "step": 3611
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.10080710443157e-08,
+      "loss": 1.6464,
+      "step": 3612
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.0747725570837394e-08,
+      "loss": 1.6263,
+      "step": 3613
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.048802890233772e-08,
+      "loss": 1.6548,
+      "step": 3614
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.02289812163168e-08,
+      "loss": 1.6505,
+      "step": 3615
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.9970582689831345e-08,
+      "loss": 1.6545,
+      "step": 3616
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.971283349949451e-08,
+      "loss": 1.663,
+      "step": 3617
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.9455733821475165e-08,
+      "loss": 1.6479,
+      "step": 3618
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.919928383149896e-08,
+      "loss": 1.6431,
+      "step": 3619
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.8943483704846465e-08,
+      "loss": 1.6274,
+      "step": 3620
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.8688333616354935e-08,
+      "loss": 1.6392,
+      "step": 3621
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.8433833740417094e-08,
+      "loss": 1.6527,
+      "step": 3622
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.8179984250981355e-08,
+      "loss": 1.63,
+      "step": 3623
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.792678532155114e-08,
+      "loss": 1.6418,
+      "step": 3624
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.7674237125185593e-08,
+      "loss": 1.625,
+      "step": 3625
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.742233983449906e-08,
+      "loss": 1.663,
+      "step": 3626
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.717109362166094e-08,
+      "loss": 1.6459,
+      "step": 3627
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.6920498658395645e-08,
+      "loss": 1.6446,
+      "step": 3628
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.667055511598272e-08,
+      "loss": 1.6275,
+      "step": 3629
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.6421263165255854e-08,
+      "loss": 1.6061,
+      "step": 3630
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.617262297660396e-08,
+      "loss": 1.6358,
+      "step": 3631
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.5924634719970215e-08,
+      "loss": 1.6476,
+      "step": 3632
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.5677298564852254e-08,
+      "loss": 1.6706,
+      "step": 3633
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.543061468030207e-08,
+      "loss": 1.631,
+      "step": 3634
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.518458323492558e-08,
+      "loss": 1.6734,
+      "step": 3635
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.4939204396883146e-08,
+      "loss": 1.6491,
+      "step": 3636
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.469447833388873e-08,
+      "loss": 1.6592,
+      "step": 3637
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.445040521321042e-08,
+      "loss": 1.657,
+      "step": 3638
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.4206985201669874e-08,
+      "loss": 1.6398,
+      "step": 3639
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.396421846564235e-08,
+      "loss": 1.6288,
+      "step": 3640
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.3722105171056566e-08,
+      "loss": 1.6399,
+      "step": 3641
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.348064548339492e-08,
+      "loss": 1.6328,
+      "step": 3642
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.3239839567692506e-08,
+      "loss": 1.6517,
+      "step": 3643
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.2999687588538117e-08,
+      "loss": 1.6624,
+      "step": 3644
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.276018971007322e-08,
+      "loss": 1.6324,
+      "step": 3645
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.252134609599245e-08,
+      "loss": 1.6642,
+      "step": 3646
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.2283156909542985e-08,
+      "loss": 1.6632,
+      "step": 3647
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.204562231352515e-08,
+      "loss": 1.6316,
+      "step": 3648
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.180874247029165e-08,
+      "loss": 1.6652,
+      "step": 3649
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.157251754174729e-08,
+      "loss": 1.6493,
+      "step": 3650
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.1336947689349924e-08,
+      "loss": 1.6636,
+      "step": 3651
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.110203307410898e-08,
+      "loss": 1.6656,
+      "step": 3652
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.0867773856586575e-08,
+      "loss": 1.6301,
+      "step": 3653
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.0634170196896746e-08,
+      "loss": 1.6404,
+      "step": 3654
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.0401222254705325e-08,
+      "loss": 1.6559,
+      "step": 3655
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.016893018922996e-08,
+      "loss": 1.6398,
+      "step": 3656
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.993729415924007e-08,
+      "loss": 1.6904,
+      "step": 3657
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.9706314323056936e-08,
+      "loss": 1.6554,
+      "step": 3658
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.947599083855291e-08,
+      "loss": 1.6615,
+      "step": 3659
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.924632386315185e-08,
+      "loss": 1.6218,
+      "step": 3660
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.90173135538292e-08,
+      "loss": 1.639,
+      "step": 3661
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.8788960067111073e-08,
+      "loss": 1.625,
+      "step": 3662
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.8561263559075075e-08,
+      "loss": 1.6504,
+      "step": 3663
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.833422418534959e-08,
+      "loss": 1.65,
+      "step": 3664
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.8107842101114064e-08,
+      "loss": 1.6522,
+      "step": 3665
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.788211746109826e-08,
+      "loss": 1.6238,
+      "step": 3666
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.765705041958289e-08,
+      "loss": 1.6731,
+      "step": 3667
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7432641130399236e-08,
+      "loss": 1.6609,
+      "step": 3668
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7208889746928926e-08,
+      "loss": 1.6408,
+      "step": 3669
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.698579642210398e-08,
+      "loss": 1.6536,
+      "step": 3670
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.6763361308406406e-08,
+      "loss": 1.655,
+      "step": 3671
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.65415845578686e-08,
+      "loss": 1.6469,
+      "step": 3672
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.632046632207286e-08,
+      "loss": 1.6555,
+      "step": 3673
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.6100006752151414e-08,
+      "loss": 1.6584,
+      "step": 3674
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.588020599878638e-08,
+      "loss": 1.6269,
+      "step": 3675
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.566106421220949e-08,
+      "loss": 1.6355,
+      "step": 3676
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.544258154220192e-08,
+      "loss": 1.6501,
+      "step": 3677
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.522475813809467e-08,
+      "loss": 1.6532,
+      "step": 3678
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.500759414876797e-08,
+      "loss": 1.6401,
+      "step": 3679
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.4791089722651435e-08,
+      "loss": 1.6459,
+      "step": 3680
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.4575245007723576e-08,
+      "loss": 1.6439,
+      "step": 3681
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.436006015151227e-08,
+      "loss": 1.6634,
+      "step": 3682
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.4145535301094545e-08,
+      "loss": 1.6629,
+      "step": 3683
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.393167060309587e-08,
+      "loss": 1.6453,
+      "step": 3684
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.371846620369101e-08,
+      "loss": 1.6424,
+      "step": 3685
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.3505922248602824e-08,
+      "loss": 1.6619,
+      "step": 3686
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.3294038883103335e-08,
+      "loss": 1.6375,
+      "step": 3687
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.308281625201292e-08,
+      "loss": 1.6303,
+      "step": 3688
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.287225449969999e-08,
+      "loss": 1.6318,
+      "step": 3689
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.266235377008175e-08,
+      "loss": 1.6572,
+      "step": 3690
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.245311420662311e-08,
+      "loss": 1.6394,
+      "step": 3691
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.224453595233756e-08,
+      "loss": 1.6388,
+      "step": 3692
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.203661914978628e-08,
+      "loss": 1.6509,
+      "step": 3693
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.182936394107838e-08,
+      "loss": 1.6593,
+      "step": 3694
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.1622770467871093e-08,
+      "loss": 1.6209,
+      "step": 3695
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.141683887136892e-08,
+      "loss": 1.6448,
+      "step": 3696
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.121156929232405e-08,
+      "loss": 1.675,
+      "step": 3697
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.1006961871036576e-08,
+      "loss": 1.6345,
+      "step": 3698
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.080301674735353e-08,
+      "loss": 1.6566,
+      "step": 3699
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.0599734060669626e-08,
+      "loss": 1.6576,
+      "step": 3700
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.039711394992661e-08,
+      "loss": 1.6679,
+      "step": 3701
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.019515655361327e-08,
+      "loss": 1.6652,
+      "step": 3702
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.999386200976573e-08,
+      "loss": 1.647,
+      "step": 3703
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.9793230455966933e-08,
+      "loss": 1.6482,
+      "step": 3704
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.9593262029346644e-08,
+      "loss": 1.6686,
+      "step": 3705
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.9393956866581304e-08,
+      "loss": 1.6704,
+      "step": 3706
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.919531510389439e-08,
+      "loss": 1.6599,
+      "step": 3707
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.899733687705519e-08,
+      "loss": 1.6658,
+      "step": 3708
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.8800022321380348e-08,
+      "loss": 1.649,
+      "step": 3709
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.8603371571732427e-08,
+      "loss": 1.6384,
+      "step": 3710
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.840738476252025e-08,
+      "loss": 1.6416,
+      "step": 3711
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.821206202769899e-08,
+      "loss": 1.6709,
+      "step": 3712
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.8017403500769864e-08,
+      "loss": 1.6279,
+      "step": 3713
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.7823409314780222e-08,
+      "loss": 1.6611,
+      "step": 3714
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.7630079602323443e-08,
+      "loss": 1.6458,
+      "step": 3715
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.7437414495538268e-08,
+      "loss": 1.6813,
+      "step": 3716
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.7245414126109478e-08,
+      "loss": 1.6634,
+      "step": 3717
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.7054078625267652e-08,
+      "loss": 1.6652,
+      "step": 3718
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6863408123788622e-08,
+      "loss": 1.6377,
+      "step": 3719
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6673402751994255e-08,
+      "loss": 1.6764,
+      "step": 3720
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6484062639751002e-08,
+      "loss": 1.667,
+      "step": 3721
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6295387916471235e-08,
+      "loss": 1.6458,
+      "step": 3722
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.610737871111246e-08,
+      "loss": 1.617,
+      "step": 3723
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5920035152176888e-08,
+      "loss": 1.6598,
+      "step": 3724
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5733357367712538e-08,
+      "loss": 1.6648,
+      "step": 3725
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5547345485311568e-08,
+      "loss": 1.6669,
+      "step": 3726
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5361999632111276e-08,
+      "loss": 1.6432,
+      "step": 3727
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5177319934793995e-08,
+      "loss": 1.6516,
+      "step": 3728
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4993306519586533e-08,
+      "loss": 1.6583,
+      "step": 3729
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.480995951226028e-08,
+      "loss": 1.6497,
+      "step": 3730
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4627279038131222e-08,
+      "loss": 1.6499,
+      "step": 3731
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4445265222059696e-08,
+      "loss": 1.6337,
+      "step": 3732
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.426391818845053e-08,
+      "loss": 1.6475,
+      "step": 3733
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4083238061252563e-08,
+      "loss": 1.6729,
+      "step": 3734
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.3903224963959133e-08,
+      "loss": 1.6721,
+      "step": 3735
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.3723879019607372e-08,
+      "loss": 1.6319,
+      "step": 3736
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.354520035077856e-08,
+      "loss": 1.6413,
+      "step": 3737
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.336718907959778e-08,
+      "loss": 1.6491,
+      "step": 3738
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.318984532773427e-08,
+      "loss": 1.6337,
+      "step": 3739
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.301316921640073e-08,
+      "loss": 1.6272,
+      "step": 3740
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.2837160866353566e-08,
+      "loss": 1.6559,
+      "step": 3741
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.2661820397892883e-08,
+      "loss": 1.6403,
+      "step": 3742
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.248714793086215e-08,
+      "loss": 1.6661,
+      "step": 3743
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.231314358464842e-08,
+      "loss": 1.6421,
+      "step": 3744
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.2139807478182004e-08,
+      "loss": 1.637,
+      "step": 3745
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.196713972993669e-08,
+      "loss": 1.6495,
+      "step": 3746
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.179514045792885e-08,
+      "loss": 1.6534,
+      "step": 3747
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.162380977971867e-08,
+      "loss": 1.6778,
+      "step": 3748
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.145314781240892e-08,
+      "loss": 1.6593,
+      "step": 3749
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.1283154672645522e-08,
+      "loss": 1.6547,
+      "step": 3750
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.111383047661719e-08,
+      "loss": 1.6313,
+      "step": 3751
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.0945175340055356e-08,
+      "loss": 1.6601,
+      "step": 3752
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.0777189378234137e-08,
+      "loss": 1.6594,
+      "step": 3753
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.06098727059707e-08,
+      "loss": 1.6384,
+      "step": 3754
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.0443225437624e-08,
+      "loss": 1.6402,
+      "step": 3755
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.0277247687096156e-08,
+      "loss": 1.6424,
+      "step": 3756
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.0111939567831193e-08,
+      "loss": 1.6855,
+      "step": 3757
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9947301192815734e-08,
+      "loss": 1.6025,
+      "step": 3758
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9783332674578547e-08,
+      "loss": 1.6246,
+      "step": 3759
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9620034125190643e-08,
+      "loss": 1.6477,
+      "step": 3760
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9457405656264968e-08,
+      "loss": 1.6488,
+      "step": 3761
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.929544737895661e-08,
+      "loss": 1.6294,
+      "step": 3762
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.913415940396246e-08,
+      "loss": 1.6442,
+      "step": 3763
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8973541841521335e-08,
+      "loss": 1.6646,
+      "step": 3764
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8813594801413756e-08,
+      "loss": 1.6445,
+      "step": 3765
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8654318392962276e-08,
+      "loss": 1.6643,
+      "step": 3766
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8495712725030477e-08,
+      "loss": 1.6416,
+      "step": 3767
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8337777906023978e-08,
+      "loss": 1.6389,
+      "step": 3768
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.818051404388976e-08,
+      "loss": 1.6551,
+      "step": 3769
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8023921246116402e-08,
+      "loss": 1.6267,
+      "step": 3770
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.7867999619733176e-08,
+      "loss": 1.6703,
+      "step": 3771
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.771274927131139e-08,
+      "loss": 1.649,
+      "step": 3772
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.755817030696294e-08,
+      "loss": 1.662,
+      "step": 3773
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.740426283234131e-08,
+      "loss": 1.6277,
+      "step": 3774
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.725102695264058e-08,
+      "loss": 1.6685,
+      "step": 3775
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.70984627725963e-08,
+      "loss": 1.6297,
+      "step": 3776
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6946570396484504e-08,
+      "loss": 1.649,
+      "step": 3777
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.679534992812215e-08,
+      "loss": 1.6239,
+      "step": 3778
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.6644801470867e-08,
+      "loss": 1.6613,
+      "step": 3779
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.6494925127617632e-08,
+      "loss": 1.6239,
+      "step": 3780
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.6345721000812996e-08,
+      "loss": 1.6653,
+      "step": 3781
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.6197189192432846e-08,
+      "loss": 1.626,
+      "step": 3782
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.6049329803997092e-08,
+      "loss": 1.63,
+      "step": 3783
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5902142936566333e-08,
+      "loss": 1.6399,
+      "step": 3784
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5755628690741428e-08,
+      "loss": 1.6462,
+      "step": 3785
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.560978716666339e-08,
+      "loss": 1.6222,
+      "step": 3786
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.546461846401359e-08,
+      "loss": 1.6529,
+      "step": 3787
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5320122682013436e-08,
+      "loss": 1.6237,
+      "step": 3788
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5176299919424485e-08,
+      "loss": 1.6339,
+      "step": 3789
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5033150274548324e-08,
+      "loss": 1.6317,
+      "step": 3790
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.489067384522613e-08,
+      "loss": 1.6346,
+      "step": 3791
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4748870728839346e-08,
+      "loss": 1.653,
+      "step": 3792
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4607741022309105e-08,
+      "loss": 1.6439,
+      "step": 3793
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4467284822096027e-08,
+      "loss": 1.6306,
+      "step": 3794
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.432750222420087e-08,
+      "loss": 1.6278,
+      "step": 3795
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.418839332416366e-08,
+      "loss": 1.6676,
+      "step": 3796
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4049958217063896e-08,
+      "loss": 1.6444,
+      "step": 3797
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.3912196997520775e-08,
+      "loss": 1.6402,
+      "step": 3798
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.377510975969265e-08,
+      "loss": 1.6658,
+      "step": 3799
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.3638696597277677e-08,
+      "loss": 1.6493,
+      "step": 3800
+    },
+    {
+      "epoch": 2.67,
+      "eval_loss": 1.653112769126892,
+      "eval_runtime": 45.3568,
+      "eval_samples_per_second": 44.095,
+      "eval_steps_per_second": 0.11,
+      "step": 3800
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.350295760351261e-08,
+      "loss": 1.6562,
+      "step": 3801
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.3367892871173902e-08,
+      "loss": 1.6425,
+      "step": 3802
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.3233502492577041e-08,
+      "loss": 1.6351,
+      "step": 3803
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.3099786559576553e-08,
+      "loss": 1.6371,
+      "step": 3804
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2966745163566106e-08,
+      "loss": 1.6189,
+      "step": 3805
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2834378395478185e-08,
+      "loss": 1.6124,
+      "step": 3806
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.2702686345784087e-08,
+      "loss": 1.6374,
+      "step": 3807
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.2571669104494253e-08,
+      "loss": 1.6417,
+      "step": 3808
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.2441326761157723e-08,
+      "loss": 1.6462,
+      "step": 3809
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.231165940486234e-08,
+      "loss": 1.6588,
+      "step": 3810
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.2182667124234324e-08,
+      "loss": 1.6582,
+      "step": 3811
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.2054350007438707e-08,
+      "loss": 1.6418,
+      "step": 3812
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.192670814217911e-08,
+      "loss": 1.6706,
+      "step": 3813
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1799741615697522e-08,
+      "loss": 1.6491,
+      "step": 3814
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.167345051477442e-08,
+      "loss": 1.6508,
+      "step": 3815
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1547834925728529e-08,
+      "loss": 1.6619,
+      "step": 3816
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1422894934416838e-08,
+      "loss": 1.6732,
+      "step": 3817
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1298630626234818e-08,
+      "loss": 1.6663,
+      "step": 3818
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.117504208611586e-08,
+      "loss": 1.6594,
+      "step": 3819
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1052129398531506e-08,
+      "loss": 1.6238,
+      "step": 3820
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0929892647491668e-08,
+      "loss": 1.6353,
+      "step": 3821
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0808331916543622e-08,
+      "loss": 1.6163,
+      "step": 3822
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0687447288773244e-08,
+      "loss": 1.6203,
+      "step": 3823
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0567238846803995e-08,
+      "loss": 1.6213,
+      "step": 3824
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0447706672797263e-08,
+      "loss": 1.647,
+      "step": 3825
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0328850848452031e-08,
+      "loss": 1.6474,
+      "step": 3826
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0210671455005204e-08,
+      "loss": 1.6242,
+      "step": 3827
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0093168573231392e-08,
+      "loss": 1.6385,
+      "step": 3828
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.976342283442462e-09,
+      "loss": 1.6804,
+      "step": 3829
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.860192665488209e-09,
+      "loss": 1.6484,
+      "step": 3830
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.744719798755908e-09,
+      "loss": 1.6573,
+      "step": 3831
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.629923762170089e-09,
+      "loss": 1.6071,
+      "step": 3832
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.515804634192658e-09,
+      "loss": 1.6425,
+      "step": 3833
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.402362492823002e-09,
+      "loss": 1.656,
+      "step": 3834
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.289597415597872e-09,
+      "loss": 1.6298,
+      "step": 3835
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.177509479591172e-09,
+      "loss": 1.6581,
+      "step": 3836
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.066098761413732e-09,
+      "loss": 1.6423,
+      "step": 3837
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.95536533721386e-09,
+      "loss": 1.6721,
+      "step": 3838
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.845309282676793e-09,
+      "loss": 1.6419,
+      "step": 3839
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.735930673024805e-09,
+      "loss": 1.6509,
+      "step": 3840
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.627229583017203e-09,
+      "loss": 1.6406,
+      "step": 3841
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.519206086950004e-09,
+      "loss": 1.6475,
+      "step": 3842
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.411860258656255e-09,
+      "loss": 1.667,
+      "step": 3843
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.305192171506047e-09,
+      "loss": 1.6641,
+      "step": 3844
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.199201898405839e-09,
+      "loss": 1.6665,
+      "step": 3845
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.093889511799013e-09,
+      "loss": 1.6471,
+      "step": 3846
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 7.989255083665658e-09,
+      "loss": 1.6307,
+      "step": 3847
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 7.885298685522235e-09,
+      "loss": 1.653,
+      "step": 3848
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 7.782020388422017e-09,
+      "loss": 1.6745,
+      "step": 3849
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 7.679420262954983e-09,
+      "loss": 1.6484,
+      "step": 3850
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 7.577498379247038e-09,
+      "loss": 1.6439,
+      "step": 3851
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 7.476254806961013e-09,
+      "loss": 1.6299,
+      "step": 3852
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 7.375689615295888e-09,
+      "loss": 1.6647,
+      "step": 3853
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 7.2758028729870135e-09,
+      "loss": 1.6437,
+      "step": 3854
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 7.176594648306111e-09,
+      "loss": 1.6273,
+      "step": 3855
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 7.07806500906094e-09,
+      "loss": 1.6548,
+      "step": 3856
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.980214022595743e-09,
+      "loss": 1.6645,
+      "step": 3857
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.883041755790575e-09,
+      "loss": 1.6622,
+      "step": 3858
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.786548275061754e-09,
+      "loss": 1.6517,
+      "step": 3859
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.690733646361857e-09,
+      "loss": 1.6475,
+      "step": 3860
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.595597935179164e-09,
+      "loss": 1.6642,
+      "step": 3861
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.501141206537886e-09,
+      "loss": 1.6339,
+      "step": 3862
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.407363524998599e-09,
+      "loss": 1.6405,
+      "step": 3863
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.3142649546572556e-09,
+      "loss": 1.6055,
+      "step": 3864
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.221845559146066e-09,
+      "loss": 1.6564,
+      "step": 3865
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.130105401632502e-09,
+      "loss": 1.6273,
+      "step": 3866
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.039044544820404e-09,
+      "loss": 1.6697,
+      "step": 3867
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.9486630509487656e-09,
+      "loss": 1.6508,
+      "step": 3868
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.8589609817925045e-09,
+      "loss": 1.6459,
+      "step": 3869
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.7699383986623554e-09,
+      "loss": 1.6556,
+      "step": 3870
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.681595362404312e-09,
+      "loss": 1.6549,
+      "step": 3871
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.593931933399853e-09,
+      "loss": 1.6334,
+      "step": 3872
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.506948171566272e-09,
+      "loss": 1.6478,
+      "step": 3873
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.420644136356123e-09,
+      "loss": 1.6784,
+      "step": 3874
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.3350198867574414e-09,
+      "loss": 1.6634,
+      "step": 3875
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.2500754812935255e-09,
+      "loss": 1.6493,
+      "step": 3876
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.165810978023044e-09,
+      "loss": 1.6463,
+      "step": 3877
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.082226434540038e-09,
+      "loss": 1.6637,
+      "step": 3878
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.999321907973697e-09,
+      "loss": 1.6626,
+      "step": 3879
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.917097454988584e-09,
+      "loss": 1.6716,
+      "step": 3880
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.835553131784298e-09,
+      "loss": 1.6606,
+      "step": 3881
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.75468899409559e-09,
+      "loss": 1.6719,
+      "step": 3882
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.67450509719236e-09,
+      "loss": 1.6258,
+      "step": 3883
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.595001495879547e-09,
+      "loss": 1.658,
+      "step": 3884
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.516178244497127e-09,
+      "loss": 1.6307,
+      "step": 3885
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.438035396920003e-09,
+      "loss": 1.6586,
+      "step": 3886
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.360573006558122e-09,
+      "loss": 1.6538,
+      "step": 3887
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.28379112635624e-09,
+      "loss": 1.6663,
+      "step": 3888
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.207689808794046e-09,
+      "loss": 1.628,
+      "step": 3889
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.132269105886154e-09,
+      "loss": 1.6205,
+      "step": 3890
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.057529069181775e-09,
+      "loss": 1.624,
+      "step": 3891
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.983469749765267e-09,
+      "loss": 1.6111,
+      "step": 3892
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9100911982552495e-09,
+      "loss": 1.6413,
+      "step": 3893
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.837393464805272e-09,
+      "loss": 1.6208,
+      "step": 3894
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.765376599103587e-09,
+      "loss": 1.6497,
+      "step": 3895
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6940406503731536e-09,
+      "loss": 1.6798,
+      "step": 3896
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6233856673713037e-09,
+      "loss": 1.6471,
+      "step": 3897
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.5534116983900743e-09,
+      "loss": 1.646,
+      "step": 3898
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.4841187912559856e-09,
+      "loss": 1.6285,
+      "step": 3899
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.415506993330153e-09,
+      "loss": 1.6534,
+      "step": 3900
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.3475763515080634e-09,
+      "loss": 1.6518,
+      "step": 3901
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.2803269122196888e-09,
+      "loss": 1.6419,
+      "step": 3902
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.2137587214293714e-09,
+      "loss": 1.6562,
+      "step": 3903
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.147871824635717e-09,
+      "loss": 1.6561,
+      "step": 3904
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.082666266872036e-09,
+      "loss": 1.6201,
+      "step": 3905
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.018142092705567e-09,
+      "loss": 1.6573,
+      "step": 3906
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.9542993462380316e-09,
+      "loss": 1.6592,
+      "step": 3907
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.8911380711051926e-09,
+      "loss": 1.662,
+      "step": 3908
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.8286583104774054e-09,
+      "loss": 1.6606,
+      "step": 3909
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.766860107058844e-09,
+      "loss": 1.6525,
+      "step": 3910
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.7057435030881648e-09,
+      "loss": 1.6397,
+      "step": 3911
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.645308540337843e-09,
+      "loss": 1.6371,
+      "step": 3912
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.5855552601146136e-09,
+      "loss": 1.6461,
+      "step": 3913
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.526483703259252e-09,
+      "loss": 1.6574,
+      "step": 3914
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.4680939101466845e-09,
+      "loss": 1.6592,
+      "step": 3915
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.4103859206857646e-09,
+      "loss": 1.6624,
+      "step": 3916
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.3533597743194965e-09,
+      "loss": 1.659,
+      "step": 3917
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2970155100244805e-09,
+      "loss": 1.6716,
+      "step": 3918
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.241353166311577e-09,
+      "loss": 1.6564,
+      "step": 3919
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.186372781225465e-09,
+      "loss": 1.6557,
+      "step": 3920
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1320743923447514e-09,
+      "loss": 1.6669,
+      "step": 3921
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.0784580367818607e-09,
+      "loss": 1.6169,
+      "step": 3922
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.0255237511830335e-09,
+      "loss": 1.6537,
+      "step": 3923
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.973271571728441e-09,
+      "loss": 1.634,
+      "step": 3924
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9217015341318475e-09,
+      "loss": 1.6798,
+      "step": 3925
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.8708136736409476e-09,
+      "loss": 1.6371,
+      "step": 3926
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.8206080250372512e-09,
+      "loss": 1.6545,
+      "step": 3927
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7710846226355324e-09,
+      "loss": 1.6492,
+      "step": 3928
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7222435002847147e-09,
+      "loss": 1.6651,
+      "step": 3929
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6740846913674279e-09,
+      "loss": 1.6337,
+      "step": 3930
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6266082287994532e-09,
+      "loss": 1.6538,
+      "step": 3931
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5798141450307222e-09,
+      "loss": 1.6552,
+      "step": 3932
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.53370247204454e-09,
+      "loss": 1.6318,
+      "step": 3933
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.488273241357696e-09,
+      "loss": 1.6478,
+      "step": 3934
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4435264840205741e-09,
+      "loss": 1.6401,
+      "step": 3935
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.3994622306173764e-09,
+      "loss": 1.6435,
+      "step": 3936
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.356080511265567e-09,
+      "loss": 1.6373,
+      "step": 3937
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.3133813556159834e-09,
+      "loss": 1.6447,
+      "step": 3938
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.27136479285328e-09,
+      "loss": 1.6658,
+      "step": 3939
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.2300308516952628e-09,
+      "loss": 1.6371,
+      "step": 3940
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.1893795603932222e-09,
+      "loss": 1.6697,
+      "step": 3941
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.1494109467321545e-09,
+      "loss": 1.6536,
+      "step": 3942
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.1101250380300965e-09,
+      "loss": 1.6438,
+      "step": 3943
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.071521861138458e-09,
+      "loss": 1.6397,
+      "step": 3944
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0336014424424667e-09,
+      "loss": 1.6333,
+      "step": 3945
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.963638078600567e-10,
+      "loss": 1.6533,
+      "step": 3946
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.598089828430911e-10,
+      "loss": 1.6752,
+      "step": 3947
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.239369923762508e-10,
+      "loss": 1.6703,
+      "step": 3948
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.887478609777011e-10,
+      "loss": 1.6592,
+      "step": 3949
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.542416126989804e-10,
+      "loss": 1.6229,
+      "step": 3950
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.204182711246676e-10,
+      "loss": 1.6532,
+      "step": 3951
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.872778593728257e-10,
+      "loss": 1.6327,
+      "step": 3952
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.548204000945579e-10,
+      "loss": 1.6625,
+      "step": 3953
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.230459154742297e-10,
+      "loss": 1.6798,
+      "step": 3954
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.919544272293576e-10,
+      "loss": 1.6637,
+      "step": 3955
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.615459566108317e-10,
+      "loss": 1.6292,
+      "step": 3956
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.3182052440236e-10,
+      "loss": 1.6632,
+      "step": 3957
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.027781509210239e-10,
+      "loss": 1.6304,
+      "step": 3958
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.74418856017167e-10,
+      "loss": 1.6511,
+      "step": 3959
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.46742659073951e-10,
+      "loss": 1.6458,
+      "step": 3960
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.19749579007911e-10,
+      "loss": 1.6408,
+      "step": 3961
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.934396342683999e-10,
+      "loss": 1.6659,
+      "step": 3962
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.678128428382555e-10,
+      "loss": 1.6609,
+      "step": 3963
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.4286922223291113e-10,
+      "loss": 1.6051,
+      "step": 3964
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.1860878950117364e-10,
+      "loss": 1.6087,
+      "step": 3965
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.9503156122489003e-10,
+      "loss": 1.6336,
+      "step": 3966
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.7213755351883647e-10,
+      "loss": 1.6407,
+      "step": 3967
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4992678203071835e-10,
+      "loss": 1.6356,
+      "step": 3968
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.2839926194161426e-10,
+      "loss": 1.6798,
+      "step": 3969
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.0755500796531e-10,
+      "loss": 1.6429,
+      "step": 3970
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.8739403434852037e-10,
+      "loss": 1.6276,
+      "step": 3971
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6791635487122265e-10,
+      "loss": 1.6548,
+      "step": 3972
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.491219828462121e-10,
+      "loss": 1.6692,
+      "step": 3973
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.310109311192132e-10,
+      "loss": 1.6436,
+      "step": 3974
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1358321206899066e-10,
+      "loss": 1.6189,
+      "step": 3975
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9683883760723829e-10,
+      "loss": 1.652,
+      "step": 3976
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.8077781917846813e-10,
+      "loss": 1.6365,
+      "step": 3977
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.654001677604544e-10,
+      "loss": 1.648,
+      "step": 3978
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.5070589386345645e-10,
+      "loss": 1.6746,
+      "step": 3979
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3669500753099582e-10,
+      "loss": 1.6654,
+      "step": 3980
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.2336751833941228e-10,
+      "loss": 1.6375,
+      "step": 3981
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.1072343539775265e-10,
+      "loss": 1.6551,
+      "step": 3982
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.8762767348326e-11,
+      "loss": 1.659,
+      "step": 3983
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.748552236603757e-11,
+      "loss": 1.633,
+      "step": 3984
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.68917081587217e-11,
+      "loss": 1.6568,
+      "step": 3985
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.698133196725297e-11,
+      "loss": 1.67,
+      "step": 3986
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.775440056521308e-11,
+      "loss": 1.6587,
+      "step": 3987
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.9210920259112884e-11,
+      "loss": 1.6348,
+      "step": 3988
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.1350896888503465e-11,
+      "loss": 1.6361,
+      "step": 3989
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.417433582542095e-11,
+      "loss": 1.6354,
+      "step": 3990
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.7681241975052728e-11,
+      "loss": 1.6696,
+      "step": 3991
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1871619775404304e-11,
+      "loss": 1.6444,
+      "step": 3992
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.674547319718833e-11,
+      "loss": 1.6472,
+      "step": 3993
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2302805744268674e-11,
+      "loss": 1.6472,
+      "step": 3994
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.543620453105305e-12,
+      "loss": 1.6414,
+      "step": 3995
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.4679198928653244e-12,
+      "loss": 1.6475,
+      "step": 3996
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.075706166089098e-12,
+      "loss": 1.6456,
+      "step": 3997
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3669809075800287e-12,
+      "loss": 1.6325,
+      "step": 3998
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.417452852927383e-13,
+      "loss": 1.6439,
+      "step": 3999
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0,
+      "loss": 1.6624,
+      "step": 4000
+    },
+    {
+      "epoch": 2.81,
+      "eval_loss": 1.6530892848968506,
+      "eval_runtime": 46.3261,
+      "eval_samples_per_second": 43.172,
+      "eval_steps_per_second": 0.108,
+      "step": 4000
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 4000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 200,
+  "total_flos": 2.51255586816e+16,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}