diff --git "a/checkpoint-1893/trainer_state.json" "b/checkpoint-1893/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1893/trainer_state.json"
@@ -0,0 +1,11411 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9997359387377872,
+  "eval_steps": 474,
+  "global_step": 1893,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 1.6827,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "eval_loss": 1.6403473615646362,
+      "eval_runtime": 682.6339,
+      "eval_samples_per_second": 9.825,
+      "eval_steps_per_second": 2.457,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-05,
+      "loss": 1.7415,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-05,
+      "loss": 1.6314,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-05,
+      "loss": 1.7016,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001,
+      "loss": 1.6104,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012,
+      "loss": 1.5584,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014,
+      "loss": 1.5481,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016,
+      "loss": 1.4103,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018,
+      "loss": 1.4447,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 1.3613,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999996538966897,
+      "loss": 1.3593,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999986155869976,
+      "loss": 1.2069,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999968850716425,
+      "loss": 1.2925,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999994462351823,
+      "loss": 1.241,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999913474292154,
+      "loss": 1.2604,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999987540305976,
+      "loss": 1.2496,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999830409847403,
+      "loss": 1.2056,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999778494686225,
+      "loss": 1.2323,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999719657612164,
+      "loss": 1.1903,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999653898665948,
+      "loss": 1.1803,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999581217893095,
+      "loss": 1.1948,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999501615343914,
+      "loss": 1.2198,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999415091073508,
+      "loss": 1.1514,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999932164514177,
+      "loss": 1.2074,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999221277613384,
+      "loss": 1.2841,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999911398855782,
+      "loss": 1.2213,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999899977804935,
+      "loss": 1.1186,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998878646167027,
+      "loss": 1.1874,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998750592994707,
+      "loss": 1.114,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998615618621018,
+      "loss": 1.1579,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000199984737231394,
+      "loss": 1.1493,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998324906648067,
+      "loss": 1.0677,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998169169250035,
+      "loss": 1.0954,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998006511053108,
+      "loss": 1.159,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997836932169875,
+      "loss": 1.1669,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999766043271772,
+      "loss": 1.2065,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997477012818822,
+      "loss": 1.1624,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997286672600138,
+      "loss": 1.1287,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997089412193426,
+      "loss": 1.0498,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019996885231735232,
+      "loss": 1.1984,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019996674131366891,
+      "loss": 1.1827,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019996456111234527,
+      "loss": 1.1191,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019996231171489056,
+      "loss": 1.1401,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019995999312286182,
+      "loss": 1.1514,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000199957605337864,
+      "loss": 1.1748,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999551483615499,
+      "loss": 1.0959,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019995262219562034,
+      "loss": 1.04,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019995002684182386,
+      "loss": 1.0853,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999473623019571,
+      "loss": 1.0513,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019994462857786431,
+      "loss": 1.1221,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999418256714379,
+      "loss": 1.1019,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019993895358461807,
+      "loss": 1.0554,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019993601231939283,
+      "loss": 1.0659,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999330018777982,
+      "loss": 1.1068,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000199929922261918,
+      "loss": 1.0432,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019992677347388394,
+      "loss": 0.9979,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999235555158757,
+      "loss": 1.0731,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019992026839012067,
+      "loss": 1.0855,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019991691209889434,
+      "loss": 1.0563,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019991348664451987,
+      "loss": 1.0464,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019990999202936841,
+      "loss": 1.0727,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019990642825585896,
+      "loss": 1.0831,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999027953264584,
+      "loss": 1.0418,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001998990932436814,
+      "loss": 1.1052,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019989532201009067,
+      "loss": 0.9511,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001998914816282966,
+      "loss": 0.9567,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019988757210095756,
+      "loss": 1.023,
+      "step": 67
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019988359343077976,
+      "loss": 1.0098,
+      "step": 68
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019987954562051725,
+      "loss": 1.0096,
+      "step": 69
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019987542867297195,
+      "loss": 0.9689,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001998712425909936,
+      "loss": 0.975,
+      "step": 71
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019986698737747992,
+      "loss": 0.9841,
+      "step": 72
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019986266303537634,
+      "loss": 0.9613,
+      "step": 73
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001998582695676762,
+      "loss": 1.0895,
+      "step": 74
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019985380697742068,
+      "loss": 0.9522,
+      "step": 75
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019984927526769884,
+      "loss": 1.0116,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001998446744416476,
+      "loss": 0.9006,
+      "step": 77
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019984000450245156,
+      "loss": 0.9101,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001998352654533434,
+      "loss": 1.0029,
+      "step": 79
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019983045729760345,
+      "loss": 0.8863,
+      "step": 80
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019982558003856,
+      "loss": 0.9774,
+      "step": 81
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019982063367958905,
+      "loss": 0.863,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019981561822411458,
+      "loss": 1.0388,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001998105336756083,
+      "loss": 0.9378,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019980538003758972,
+      "loss": 1.052,
+      "step": 85
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019980015731362627,
+      "loss": 0.9523,
+      "step": 86
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019979486550733314,
+      "loss": 0.929,
+      "step": 87
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019978950462237337,
+      "loss": 0.9387,
+      "step": 88
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019978407466245777,
+      "loss": 0.9565,
+      "step": 89
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000199778575631345,
+      "loss": 1.0259,
+      "step": 90
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019977300753284158,
+      "loss": 0.9605,
+      "step": 91
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997673703708017,
+      "loss": 0.8683,
+      "step": 92
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997616641491275,
+      "loss": 0.8937,
+      "step": 93
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019975588887176883,
+      "loss": 0.8495,
+      "step": 94
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019975004454272338,
+      "loss": 0.8717,
+      "step": 95
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997441311660367,
+      "loss": 0.9684,
+      "step": 96
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019973814874580198,
+      "loss": 0.879,
+      "step": 97
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019973209728616032,
+      "loss": 0.927,
+      "step": 98
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019972597679130058,
+      "loss": 1.0419,
+      "step": 99
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019971978726545942,
+      "loss": 0.9943,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019971352871292125,
+      "loss": 0.8513,
+      "step": 101
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019970720113801827,
+      "loss": 0.9358,
+      "step": 102
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019970080454513055,
+      "loss": 0.8453,
+      "step": 103
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019969433893868578,
+      "loss": 0.8896,
+      "step": 104
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996878043231595,
+      "loss": 0.8942,
+      "step": 105
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000199681200703075,
+      "loss": 0.9442,
+      "step": 106
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019967452808300343,
+      "loss": 0.8504,
+      "step": 107
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019966778646756354,
+      "loss": 0.8818,
+      "step": 108
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019966097586142194,
+      "loss": 0.92,
+      "step": 109
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000199654096269293,
+      "loss": 0.9062,
+      "step": 110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019964714769593882,
+      "loss": 0.9225,
+      "step": 111
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996401301461692,
+      "loss": 0.8717,
+      "step": 112
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019963304362484178,
+      "loss": 0.8728,
+      "step": 113
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019962588813686187,
+      "loss": 0.8795,
+      "step": 114
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996186636871826,
+      "loss": 0.8862,
+      "step": 115
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996113702808047,
+      "loss": 0.9683,
+      "step": 116
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019960400792277678,
+      "loss": 0.876,
+      "step": 117
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019959657661819508,
+      "loss": 0.8434,
+      "step": 118
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995890763722036,
+      "loss": 0.8244,
+      "step": 119
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019958150718999408,
+      "loss": 0.8203,
+      "step": 120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019957386907680594,
+      "loss": 0.8332,
+      "step": 121
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019956616203792635,
+      "loss": 0.8765,
+      "step": 122
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019955838607869014,
+      "loss": 0.8099,
+      "step": 123
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019955054120447994,
+      "loss": 0.7839,
+      "step": 124
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019954262742072592,
+      "loss": 0.8986,
+      "step": 125
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001995346447329062,
+      "loss": 0.8446,
+      "step": 126
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001995265931465463,
+      "loss": 0.8019,
+      "step": 127
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001995184726672197,
+      "loss": 0.7754,
+      "step": 128
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019951028330054737,
+      "loss": 0.8713,
+      "step": 129
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019950202505219808,
+      "loss": 0.8419,
+      "step": 130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019949369792788825,
+      "loss": 0.8456,
+      "step": 131
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019948530193338195,
+      "loss": 0.7685,
+      "step": 132
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019947683707449094,
+      "loss": 0.8374,
+      "step": 133
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019946830335707468,
+      "loss": 0.8055,
+      "step": 134
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019945970078704024,
+      "loss": 0.809,
+      "step": 135
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019945102937034238,
+      "loss": 0.8346,
+      "step": 136
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019944228911298353,
+      "loss": 0.8243,
+      "step": 137
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019943348002101371,
+      "loss": 0.6564,
+      "step": 138
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001994246021005307,
+      "loss": 0.8963,
+      "step": 139
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019941565535767981,
+      "loss": 0.7951,
+      "step": 140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000199406639798654,
+      "loss": 0.6793,
+      "step": 141
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000199397555429694,
+      "loss": 0.8094,
+      "step": 142
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000199388402257088,
+      "loss": 0.9244,
+      "step": 143
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019937918028717187,
+      "loss": 0.836,
+      "step": 144
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019936988952632916,
+      "loss": 0.7727,
+      "step": 145
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000199360529980991,
+      "loss": 0.7911,
+      "step": 146
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001993511016576361,
+      "loss": 0.8511,
+      "step": 147
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001993416045627908,
+      "loss": 0.8016,
+      "step": 148
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019933203870302907,
+      "loss": 0.8169,
+      "step": 149
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001993224040849725,
+      "loss": 0.7972,
+      "step": 150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019931270071529017,
+      "loss": 0.7496,
+      "step": 151
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019930292860069888,
+      "loss": 0.8504,
+      "step": 152
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001992930877479629,
+      "loss": 0.7957,
+      "step": 153
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019928317816389417,
+      "loss": 0.8374,
+      "step": 154
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019927319985535213,
+      "loss": 0.7379,
+      "step": 155
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019926315282924387,
+      "loss": 0.7342,
+      "step": 156
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000199253037092524,
+      "loss": 0.6978,
+      "step": 157
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019924285265219467,
+      "loss": 0.7789,
+      "step": 158
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019923259951530565,
+      "loss": 0.7688,
+      "step": 159
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001992222776889542,
+      "loss": 0.8388,
+      "step": 160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019921188718028518,
+      "loss": 0.7624,
+      "step": 161
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019920142799649098,
+      "loss": 0.8582,
+      "step": 162
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001991909001448115,
+      "loss": 0.9053,
+      "step": 163
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019918030363253418,
+      "loss": 0.6999,
+      "step": 164
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019916963846699403,
+      "loss": 0.7555,
+      "step": 165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019915890465557347,
+      "loss": 0.783,
+      "step": 166
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001991481022057026,
+      "loss": 0.7885,
+      "step": 167
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001991372311248589,
+      "loss": 0.6863,
+      "step": 168
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019912629142056744,
+      "loss": 0.8085,
+      "step": 169
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019911528310040074,
+      "loss": 0.769,
+      "step": 170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001991042061719788,
+      "loss": 0.7847,
+      "step": 171
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001990930606429692,
+      "loss": 0.7799,
+      "step": 172
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001990818465210869,
+      "loss": 0.7755,
+      "step": 173
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001990705638140944,
+      "loss": 0.918,
+      "step": 174
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001990592125298017,
+      "loss": 0.7651,
+      "step": 175
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001990477926760662,
+      "loss": 0.6945,
+      "step": 176
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019903630426079279,
+      "loss": 0.7033,
+      "step": 177
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019902474729193386,
+      "loss": 0.7857,
+      "step": 178
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001990131217774892,
+      "loss": 0.7461,
+      "step": 179
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019900142772550603,
+      "loss": 0.7372,
+      "step": 180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019898966514407914,
+      "loss": 0.8455,
+      "step": 181
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001989778340413506,
+      "loss": 0.7223,
+      "step": 182
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019896593442551,
+      "loss": 0.6334,
+      "step": 183
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019895396630479435,
+      "loss": 0.6756,
+      "step": 184
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019894192968748802,
+      "loss": 0.7634,
+      "step": 185
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019892982458192288,
+      "loss": 0.6425,
+      "step": 186
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001989176509964781,
+      "loss": 0.7796,
+      "step": 187
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001989054089395804,
+      "loss": 0.7484,
+      "step": 188
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019889309841970374,
+      "loss": 0.7369,
+      "step": 189
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001988807194453696,
+      "loss": 0.8663,
+      "step": 190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019886827202514673,
+      "loss": 0.6501,
+      "step": 191
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019885575616765137,
+      "loss": 0.6236,
+      "step": 192
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019884317188154706,
+      "loss": 0.7812,
+      "step": 193
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019883051917554472,
+      "loss": 0.6338,
+      "step": 194
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019881779805840265,
+      "loss": 0.6472,
+      "step": 195
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019880500853892646,
+      "loss": 0.7427,
+      "step": 196
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001987921506259692,
+      "loss": 0.7296,
+      "step": 197
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019877922432843116,
+      "loss": 0.6716,
+      "step": 198
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019876622965525997,
+      "loss": 0.658,
+      "step": 199
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001987531666154507,
+      "loss": 0.676,
+      "step": 200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019874003521804564,
+      "loss": 0.6148,
+      "step": 201
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019872683547213446,
+      "loss": 0.6585,
+      "step": 202
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001987135673868541,
+      "loss": 0.6368,
+      "step": 203
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019870023097138875,
+      "loss": 0.716,
+      "step": 204
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019868682623497006,
+      "loss": 0.7773,
+      "step": 205
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001986733531868768,
+      "loss": 0.737,
+      "step": 206
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019865981183643516,
+      "loss": 0.7895,
+      "step": 207
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019864620219301856,
+      "loss": 0.7056,
+      "step": 208
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001986325242660476,
+      "loss": 0.798,
+      "step": 209
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019861877806499035,
+      "loss": 0.675,
+      "step": 210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001986049635993619,
+      "loss": 0.7328,
+      "step": 211
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019859108087872483,
+      "loss": 0.5785,
+      "step": 212
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019857712991268875,
+      "loss": 0.6947,
+      "step": 213
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001985631107109107,
+      "loss": 0.6767,
+      "step": 214
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019854902328309478,
+      "loss": 0.6632,
+      "step": 215
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019853486763899247,
+      "loss": 0.6893,
+      "step": 216
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019852064378840235,
+      "loss": 0.7121,
+      "step": 217
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019850635174117033,
+      "loss": 0.6771,
+      "step": 218
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019849199150718936,
+      "loss": 0.6797,
+      "step": 219
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001984775630963998,
+      "loss": 0.7419,
+      "step": 220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000198463066518789,
+      "loss": 0.8314,
+      "step": 221
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019844850178439162,
+      "loss": 0.6751,
+      "step": 222
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001984338689032895,
+      "loss": 0.6672,
+      "step": 223
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019841916788561155,
+      "loss": 0.583,
+      "step": 224
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019840439874153397,
+      "loss": 0.6423,
+      "step": 225
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019838956148128002,
+      "loss": 0.6227,
+      "step": 226
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019837465611512018,
+      "loss": 0.6453,
+      "step": 227
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019835968265337204,
+      "loss": 0.7909,
+      "step": 228
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019834464110640032,
+      "loss": 0.6484,
+      "step": 229
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019832953148461683,
+      "loss": 0.6466,
+      "step": 230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019831435379848066,
+      "loss": 0.6341,
+      "step": 231
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001982991080584978,
+      "loss": 0.7028,
+      "step": 232
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019828379427522155,
+      "loss": 0.6312,
+      "step": 233
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019826841245925212,
+      "loss": 0.6302,
+      "step": 234
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019825296262123692,
+      "loss": 0.5905,
+      "step": 235
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001982374447718705,
+      "loss": 0.5553,
+      "step": 236
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019822185892189433,
+      "loss": 0.7098,
+      "step": 237
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019820620508209712,
+      "loss": 0.5711,
+      "step": 238
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019819048326331448,
+      "loss": 0.5917,
+      "step": 239
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001981746934764292,
+      "loss": 0.7259,
+      "step": 240
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001981588357323711,
+      "loss": 0.6311,
+      "step": 241
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019814291004211697,
+      "loss": 0.5428,
+      "step": 242
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001981269164166907,
+      "loss": 0.6518,
+      "step": 243
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019811085486716318,
+      "loss": 0.5686,
+      "step": 244
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019809472540465234,
+      "loss": 0.652,
+      "step": 245
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019807852804032305,
+      "loss": 0.6471,
+      "step": 246
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001980622627853873,
+      "loss": 0.5876,
+      "step": 247
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019804592965110395,
+      "loss": 0.4256,
+      "step": 248
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001980295286487789,
+      "loss": 0.4416,
+      "step": 249
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001980130597897651,
+      "loss": 0.6053,
+      "step": 250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019799652308546235,
+      "loss": 0.6321,
+      "step": 251
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019797991854731745,
+      "loss": 0.6391,
+      "step": 252
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019796324618682424,
+      "loss": 0.5523,
+      "step": 253
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019794650601552335,
+      "loss": 0.551,
+      "step": 254
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019792969804500253,
+      "loss": 0.6906,
+      "step": 255
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001979128222868963,
+      "loss": 0.622,
+      "step": 256
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019789587875288617,
+      "loss": 0.5587,
+      "step": 257
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019787886745470065,
+      "loss": 0.5844,
+      "step": 258
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019786178840411496,
+      "loss": 0.511,
+      "step": 259
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019784464161295143,
+      "loss": 0.4971,
+      "step": 260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001978274270930791,
+      "loss": 0.4905,
+      "step": 261
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019781014485641403,
+      "loss": 0.678,
+      "step": 262
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019779279491491907,
+      "loss": 0.5679,
+      "step": 263
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000197775377280604,
+      "loss": 0.5433,
+      "step": 264
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019775789196552538,
+      "loss": 0.6029,
+      "step": 265
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019774033898178667,
+      "loss": 0.4002,
+      "step": 266
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019772271834153816,
+      "loss": 0.571,
+      "step": 267
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000197705030056977,
+      "loss": 0.5115,
+      "step": 268
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019768727414034716,
+      "loss": 0.512,
+      "step": 269
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019766945060393934,
+      "loss": 0.4941,
+      "step": 270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001976515594600911,
+      "loss": 0.6701,
+      "step": 271
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001976336007211869,
+      "loss": 0.6203,
+      "step": 272
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019761557439965777,
+      "loss": 0.5377,
+      "step": 273
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019759748050798178,
+      "loss": 0.398,
+      "step": 274
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019757931905868355,
+      "loss": 0.5649,
+      "step": 275
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019756109006433459,
+      "loss": 0.566,
+      "step": 276
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019754279353755312,
+      "loss": 0.4631,
+      "step": 277
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019752442949100415,
+      "loss": 0.5042,
+      "step": 278
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019750599793739936,
+      "loss": 0.6321,
+      "step": 279
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001974874988894972,
+      "loss": 0.6155,
+      "step": 280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019746893236010282,
+      "loss": 0.6981,
+      "step": 281
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019745029836206813,
+      "loss": 0.5124,
+      "step": 282
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001974315969082917,
+      "loss": 0.5185,
+      "step": 283
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019741282801171873,
+      "loss": 0.6237,
+      "step": 284
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001973939916853413,
+      "loss": 0.6814,
+      "step": 285
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019737508794219792,
+      "loss": 0.7321,
+      "step": 286
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019735611679537395,
+      "loss": 0.4718,
+      "step": 287
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019733707825800135,
+      "loss": 0.5707,
+      "step": 288
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001973179723432587,
+      "loss": 0.6378,
+      "step": 289
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019729879906437122,
+      "loss": 0.6632,
+      "step": 290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019727955843461086,
+      "loss": 0.5456,
+      "step": 291
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000197260250467296,
+      "loss": 0.6076,
+      "step": 292
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001972408751757918,
+      "loss": 0.5761,
+      "step": 293
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019722143257351,
+      "loss": 0.3927,
+      "step": 294
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019720192267390883,
+      "loss": 0.6148,
+      "step": 295
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001971823454904932,
+      "loss": 0.5988,
+      "step": 296
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001971627010368146,
+      "loss": 0.4857,
+      "step": 297
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019714298932647098,
+      "loss": 0.6357,
+      "step": 298
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000197123210373107,
+      "loss": 0.5416,
+      "step": 299
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019710336419041373,
+      "loss": 0.4784,
+      "step": 300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019708345079212883,
+      "loss": 0.5227,
+      "step": 301
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001970634701920365,
+      "loss": 0.6115,
+      "step": 302
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001970434224039674,
+      "loss": 0.4916,
+      "step": 303
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019702330744179883,
+      "loss": 0.4358,
+      "step": 304
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019700312531945442,
+      "loss": 0.7045,
+      "step": 305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001969828760509044,
+      "loss": 0.4024,
+      "step": 306
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019696255965016545,
+      "loss": 0.6041,
+      "step": 307
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001969421761313007,
+      "loss": 0.6045,
+      "step": 308
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019692172550841977,
+      "loss": 0.5376,
+      "step": 309
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019690120779567875,
+      "loss": 0.5746,
+      "step": 310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019688062300728007,
+      "loss": 0.4804,
+      "step": 311
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001968599711574727,
+      "loss": 0.7026,
+      "step": 312
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000196839252260552,
+      "loss": 0.5828,
+      "step": 313
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019681846633085967,
+      "loss": 0.6048,
+      "step": 314
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019679761338278393,
+      "loss": 0.4623,
+      "step": 315
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019677669343075928,
+      "loss": 0.6218,
+      "step": 316
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019675570648926667,
+      "loss": 0.5433,
+      "step": 317
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019673465257283342,
+      "loss": 0.6229,
+      "step": 318
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019671353169603315,
+      "loss": 0.3385,
+      "step": 319
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001966923438734859,
+      "loss": 0.4531,
+      "step": 320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019667108911985802,
+      "loss": 0.4682,
+      "step": 321
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001966497674498622,
+      "loss": 0.4215,
+      "step": 322
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019662837887825737,
+      "loss": 0.4271,
+      "step": 323
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001966069234198489,
+      "loss": 0.5211,
+      "step": 324
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019658540108948843,
+      "loss": 0.6122,
+      "step": 325
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019656381190207386,
+      "loss": 0.5395,
+      "step": 326
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019654215587254926,
+      "loss": 0.4182,
+      "step": 327
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019652043301590522,
+      "loss": 0.4397,
+      "step": 328
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019649864334717835,
+      "loss": 0.5954,
+      "step": 329
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001964767868814516,
+      "loss": 0.5507,
+      "step": 330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019645486363385422,
+      "loss": 0.5757,
+      "step": 331
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019643287361956159,
+      "loss": 0.4473,
+      "step": 332
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019641081685379535,
+      "loss": 0.4888,
+      "step": 333
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019638869335182336,
+      "loss": 0.5083,
+      "step": 334
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001963665031289596,
+      "loss": 0.5729,
+      "step": 335
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019634424620056432,
+      "loss": 0.4122,
+      "step": 336
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019632192258204394,
+      "loss": 0.4328,
+      "step": 337
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019629953228885098,
+      "loss": 0.3844,
+      "step": 338
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019627707533648417,
+      "loss": 0.5123,
+      "step": 339
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019625455174048835,
+      "loss": 0.3861,
+      "step": 340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001962319615164545,
+      "loss": 0.5526,
+      "step": 341
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019620930468001974,
+      "loss": 0.5327,
+      "step": 342
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019618658124686728,
+      "loss": 0.4724,
+      "step": 343
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001961637912327264,
+      "loss": 0.4352,
+      "step": 344
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019614093465337255,
+      "loss": 0.5487,
+      "step": 345
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019611801152462715,
+      "loss": 0.517,
+      "step": 346
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019609502186235776,
+      "loss": 0.6051,
+      "step": 347
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019607196568247802,
+      "loss": 0.4616,
+      "step": 348
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001960488430009475,
+      "loss": 0.3068,
+      "step": 349
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001960256538337719,
+      "loss": 0.615,
+      "step": 350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019600239819700296,
+      "loss": 0.4958,
+      "step": 351
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019597907610673833,
+      "loss": 0.6533,
+      "step": 352
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019595568757912172,
+      "loss": 0.6606,
+      "step": 353
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019593223263034287,
+      "loss": 0.4331,
+      "step": 354
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001959087112766374,
+      "loss": 0.4732,
+      "step": 355
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019588512353428694,
+      "loss": 0.5235,
+      "step": 356
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019586146941961912,
+      "loss": 0.4875,
+      "step": 357
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019583774894900747,
+      "loss": 0.4619,
+      "step": 358
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019581396213887145,
+      "loss": 0.4515,
+      "step": 359
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019579010900567644,
+      "loss": 0.4014,
+      "step": 360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019576618956593372,
+      "loss": 0.6054,
+      "step": 361
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019574220383620055,
+      "loss": 0.463,
+      "step": 362
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019571815183307996,
+      "loss": 0.5436,
+      "step": 363
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001956940335732209,
+      "loss": 0.4796,
+      "step": 364
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001956698490733182,
+      "loss": 0.5448,
+      "step": 365
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019564559835011253,
+      "loss": 0.4491,
+      "step": 366
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019562128142039043,
+      "loss": 0.4948,
+      "step": 367
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001955968983009842,
+      "loss": 0.5413,
+      "step": 368
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019557244900877198,
+      "loss": 0.5024,
+      "step": 369
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001955479335606778,
+      "loss": 0.5393,
+      "step": 370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019552335197367136,
+      "loss": 0.4885,
+      "step": 371
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019549870426476825,
+      "loss": 0.4767,
+      "step": 372
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001954739904510297,
+      "loss": 0.5413,
+      "step": 373
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019544921054956286,
+      "loss": 0.4429,
+      "step": 374
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019542436457752044,
+      "loss": 0.5065,
+      "step": 375
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001953994525521011,
+      "loss": 0.4118,
+      "step": 376
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019537447449054901,
+      "loss": 0.4231,
+      "step": 377
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019534943041015423,
+      "loss": 0.5443,
+      "step": 378
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001953243203282524,
+      "loss": 0.4195,
+      "step": 379
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001952991442622249,
+      "loss": 0.4849,
+      "step": 380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019527390222949875,
+      "loss": 0.3906,
+      "step": 381
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019524859424754665,
+      "loss": 0.5446,
+      "step": 382
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019522322033388697,
+      "loss": 0.536,
+      "step": 383
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019519778050608368,
+      "loss": 0.5869,
+      "step": 384
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019517227478174642,
+      "loss": 0.339,
+      "step": 385
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019514670317853042,
+      "loss": 0.3948,
+      "step": 386
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001951210657141365,
+      "loss": 0.3832,
+      "step": 387
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019509536240631107,
+      "loss": 0.4871,
+      "step": 388
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019506959327284619,
+      "loss": 0.5002,
+      "step": 389
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019504375833157936,
+      "loss": 0.4643,
+      "step": 390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019501785760039371,
+      "loss": 0.4528,
+      "step": 391
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019499189109721795,
+      "loss": 0.5932,
+      "step": 392
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019496585884002616,
+      "loss": 0.3793,
+      "step": 393
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019493976084683813,
+      "loss": 0.3792,
+      "step": 394
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019491359713571903,
+      "loss": 0.4917,
+      "step": 395
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019488736772477961,
+      "loss": 0.3786,
+      "step": 396
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019486107263217594,
+      "loss": 0.3366,
+      "step": 397
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019483471187610972,
+      "loss": 0.4563,
+      "step": 398
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019480828547482806,
+      "loss": 0.4573,
+      "step": 399
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019478179344662344,
+      "loss": 0.3949,
+      "step": 400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019475523580983383,
+      "loss": 0.5698,
+      "step": 401
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001947286125828426,
+      "loss": 0.3103,
+      "step": 402
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001947019237840786,
+      "loss": 0.4051,
+      "step": 403
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019467516943201586,
+      "loss": 0.4425,
+      "step": 404
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019464834954517398,
+      "loss": 0.4839,
+      "step": 405
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019462146414211793,
+      "loss": 0.491,
+      "step": 406
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019459451324145785,
+      "loss": 0.5769,
+      "step": 407
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019456749686184944,
+      "loss": 0.4706,
+      "step": 408
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019454041502199351,
+      "loss": 0.3509,
+      "step": 409
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019451326774063636,
+      "loss": 0.3738,
+      "step": 410
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019448605503656952,
+      "loss": 0.4354,
+      "step": 411
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019445877692862975,
+      "loss": 0.4723,
+      "step": 412
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019443143343569918,
+      "loss": 0.313,
+      "step": 413
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019440402457670514,
+      "loss": 0.4611,
+      "step": 414
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019437655037062026,
+      "loss": 0.2965,
+      "step": 415
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019434901083646233,
+      "loss": 0.4249,
+      "step": 416
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019432140599329442,
+      "loss": 0.3893,
+      "step": 417
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019429373586022473,
+      "loss": 0.5318,
+      "step": 418
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019426600045640676,
+      "loss": 0.4963,
+      "step": 419
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019423819980103914,
+      "loss": 0.4694,
+      "step": 420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019421033391336568,
+      "loss": 0.414,
+      "step": 421
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001941824028126753,
+      "loss": 0.5468,
+      "step": 422
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019415440651830208,
+      "loss": 0.4955,
+      "step": 423
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001941263450496253,
+      "loss": 0.4204,
+      "step": 424
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019409821842606922,
+      "loss": 0.3463,
+      "step": 425
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019407002666710336,
+      "loss": 0.4911,
+      "step": 426
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019404176979224212,
+      "loss": 0.393,
+      "step": 427
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019401344782104523,
+      "loss": 0.5595,
+      "step": 428
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019398506077311729,
+      "loss": 0.3889,
+      "step": 429
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019395660866810798,
+      "loss": 0.3107,
+      "step": 430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019392809152571206,
+      "loss": 0.3679,
+      "step": 431
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001938995093656693,
+      "loss": 0.4218,
+      "step": 432
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019387086220776443,
+      "loss": 0.4102,
+      "step": 433
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019384215007182718,
+      "loss": 0.4553,
+      "step": 434
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019381337297773234,
+      "loss": 0.3795,
+      "step": 435
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019378453094539956,
+      "loss": 0.2866,
+      "step": 436
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019375562399479354,
+      "loss": 0.4198,
+      "step": 437
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019372665214592382,
+      "loss": 0.3574,
+      "step": 438
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019369761541884487,
+      "loss": 0.3776,
+      "step": 439
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001936685138336562,
+      "loss": 0.3728,
+      "step": 440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019363934741050204,
+      "loss": 0.5994,
+      "step": 441
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019361011616957164,
+      "loss": 0.4657,
+      "step": 442
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000193580820131099,
+      "loss": 0.5495,
+      "step": 443
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019355145931536304,
+      "loss": 0.4567,
+      "step": 444
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019352203374268755,
+      "loss": 0.4354,
+      "step": 445
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001934925434334411,
+      "loss": 0.4938,
+      "step": 446
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019346298840803704,
+      "loss": 0.3425,
+      "step": 447
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001934333686869336,
+      "loss": 0.3021,
+      "step": 448
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001934036842906337,
+      "loss": 0.2868,
+      "step": 449
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001933739352396851,
+      "loss": 0.5395,
+      "step": 450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001933441215546803,
+      "loss": 0.3963,
+      "step": 451
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019331424325625657,
+      "loss": 0.4274,
+      "step": 452
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019328430036509575,
+      "loss": 0.4338,
+      "step": 453
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019325429290192457,
+      "loss": 0.4878,
+      "step": 454
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019322422088751445,
+      "loss": 0.3266,
+      "step": 455
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019319408434268132,
+      "loss": 0.2842,
+      "step": 456
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019316388328828604,
+      "loss": 0.2113,
+      "step": 457
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019313361774523385,
+      "loss": 0.4431,
+      "step": 458
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001931032877344748,
+      "loss": 0.4374,
+      "step": 459
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019307289327700358,
+      "loss": 0.3236,
+      "step": 460
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019304243439385938,
+      "loss": 0.5163,
+      "step": 461
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019301191110612605,
+      "loss": 0.4897,
+      "step": 462
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000192981323434932,
+      "loss": 0.3545,
+      "step": 463
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019295067140145022,
+      "loss": 0.3728,
+      "step": 464
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001929199550268983,
+      "loss": 0.2153,
+      "step": 465
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019288917433253825,
+      "loss": 0.258,
+      "step": 466
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019285832933967672,
+      "loss": 0.3084,
+      "step": 467
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019282742006966478,
+      "loss": 0.4221,
+      "step": 468
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019279644654389807,
+      "loss": 0.4629,
+      "step": 469
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019276540878381663,
+      "loss": 0.4955,
+      "step": 470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000192734306810905,
+      "loss": 0.2758,
+      "step": 471
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019270314064669222,
+      "loss": 0.2728,
+      "step": 472
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019267191031275172,
+      "loss": 0.3211,
+      "step": 473
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019264061583070127,
+      "loss": 0.4101,
+      "step": 474
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 0.37793055176734924,
+      "eval_runtime": 683.4711,
+      "eval_samples_per_second": 9.813,
+      "eval_steps_per_second": 2.454,
+      "step": 474
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019260925722220317,
+      "loss": 0.3679,
+      "step": 475
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019257783450896406,
+      "loss": 0.4074,
+      "step": 476
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019254634771273492,
+      "loss": 0.4984,
+      "step": 477
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019251479685531114,
+      "loss": 0.3992,
+      "step": 478
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019248318195853244,
+      "loss": 0.4017,
+      "step": 479
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019245150304428283,
+      "loss": 0.3794,
+      "step": 480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001924197601344907,
+      "loss": 0.369,
+      "step": 481
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001923879532511287,
+      "loss": 0.4137,
+      "step": 482
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001923560824162137,
+      "loss": 0.4004,
+      "step": 483
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000192324147651807,
+      "loss": 0.2607,
+      "step": 484
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000192292148980014,
+      "loss": 0.4933,
+      "step": 485
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019226008642298442,
+      "loss": 0.3053,
+      "step": 486
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019222796000291213,
+      "loss": 0.4018,
+      "step": 487
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019219576974203526,
+      "loss": 0.4102,
+      "step": 488
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019216351566263618,
+      "loss": 0.4241,
+      "step": 489
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019213119778704128,
+      "loss": 0.4772,
+      "step": 490
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019209881613762133,
+      "loss": 0.3709,
+      "step": 491
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019206637073679103,
+      "loss": 0.2739,
+      "step": 492
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001920338616070093,
+      "loss": 0.4537,
+      "step": 493
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019200128877077925,
+      "loss": 0.3589,
+      "step": 494
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019196865225064793,
+      "loss": 0.3767,
+      "step": 495
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001919359520692066,
+      "loss": 0.3263,
+      "step": 496
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019190318824909052,
+      "loss": 0.2835,
+      "step": 497
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019187036081297907,
+      "loss": 0.4297,
+      "step": 498
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019183746978359553,
+      "loss": 0.4173,
+      "step": 499
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001918045151837074,
+      "loss": 0.4225,
+      "step": 500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000191771497036126,
+      "loss": 0.5362,
+      "step": 501
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001917384153637067,
+      "loss": 0.341,
+      "step": 502
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019170527018934886,
+      "loss": 0.3225,
+      "step": 503
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019167206153599585,
+      "loss": 0.4146,
+      "step": 504
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001916387894266349,
+      "loss": 0.3541,
+      "step": 505
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019160545388429708,
+      "loss": 0.432,
+      "step": 506
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019157205493205762,
+      "loss": 0.3055,
+      "step": 507
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001915385925930354,
+      "loss": 0.3346,
+      "step": 508
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001915050668903933,
+      "loss": 0.5637,
+      "step": 509
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019147147784733806,
+      "loss": 0.2844,
+      "step": 510
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001914378254871202,
+      "loss": 0.5342,
+      "step": 511
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001914041098330341,
+      "loss": 0.3227,
+      "step": 512
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000191370330908418,
+      "loss": 0.5065,
+      "step": 513
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019133648873665392,
+      "loss": 0.5285,
+      "step": 514
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019130258334116752,
+      "loss": 0.308,
+      "step": 515
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019126861474542844,
+      "loss": 0.32,
+      "step": 516
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019123458297294992,
+      "loss": 0.3941,
+      "step": 517
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000191200488047289,
+      "loss": 0.3382,
+      "step": 518
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001911663299920464,
+      "loss": 0.3564,
+      "step": 519
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019113210883086657,
+      "loss": 0.3528,
+      "step": 520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019109782458743763,
+      "loss": 0.3915,
+      "step": 521
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019106347728549135,
+      "loss": 0.3346,
+      "step": 522
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001910290669488031,
+      "loss": 0.2841,
+      "step": 523
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019099459360119204,
+      "loss": 0.465,
+      "step": 524
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001909600572665208,
+      "loss": 0.3728,
+      "step": 525
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019092545796869565,
+      "loss": 0.4835,
+      "step": 526
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001908907957316665,
+      "loss": 0.4034,
+      "step": 527
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001908560705794267,
+      "loss": 0.3452,
+      "step": 528
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019082128253601334,
+      "loss": 0.2769,
+      "step": 529
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019078643162550685,
+      "loss": 0.3648,
+      "step": 530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019075151787203125,
+      "loss": 0.3377,
+      "step": 531
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019071654129975416,
+      "loss": 0.4024,
+      "step": 532
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001906815019328865,
+      "loss": 0.3533,
+      "step": 533
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001906463997956828,
+      "loss": 0.453,
+      "step": 534
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000190611234912441,
+      "loss": 0.3315,
+      "step": 535
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019057600730750242,
+      "loss": 0.3055,
+      "step": 536
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001905407170052519,
+      "loss": 0.293,
+      "step": 537
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001905053640301176,
+      "loss": 0.3877,
+      "step": 538
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019046994840657105,
+      "loss": 0.4629,
+      "step": 539
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019043447015912722,
+      "loss": 0.4711,
+      "step": 540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019039892931234435,
+      "loss": 0.3827,
+      "step": 541
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019036332589082407,
+      "loss": 0.2052,
+      "step": 542
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019032765991921133,
+      "loss": 0.3286,
+      "step": 543
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001902919314221943,
+      "loss": 0.3437,
+      "step": 544
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019025614042450455,
+      "loss": 0.4702,
+      "step": 545
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019022028695091675,
+      "loss": 0.3217,
+      "step": 546
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000190184371026249,
+      "loss": 0.3117,
+      "step": 547
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001901483926753625,
+      "loss": 0.4213,
+      "step": 548
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001901123519231617,
+      "loss": 0.2494,
+      "step": 549
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001900762487945943,
+      "loss": 0.2425,
+      "step": 550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019004008331465104,
+      "loss": 0.1523,
+      "step": 551
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019000385550836595,
+      "loss": 0.2795,
+      "step": 552
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018996756540081618,
+      "loss": 0.3282,
+      "step": 553
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018993121301712193,
+      "loss": 0.3516,
+      "step": 554
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001898947983824466,
+      "loss": 0.4524,
+      "step": 555
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018985832152199663,
+      "loss": 0.3308,
+      "step": 556
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018982178246102152,
+      "loss": 0.3367,
+      "step": 557
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001897851812248139,
+      "loss": 0.465,
+      "step": 558
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018974851783870936,
+      "loss": 0.2814,
+      "step": 559
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018971179232808654,
+      "loss": 0.2469,
+      "step": 560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018967500471836707,
+      "loss": 0.3493,
+      "step": 561
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001896381550350156,
+      "loss": 0.2774,
+      "step": 562
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018960124330353969,
+      "loss": 0.4549,
+      "step": 563
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001895642695494899,
+      "loss": 0.1397,
+      "step": 564
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018952723379845975,
+      "loss": 0.4118,
+      "step": 565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001894901360760856,
+      "loss": 0.5004,
+      "step": 566
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018945297640804673,
+      "loss": 0.3488,
+      "step": 567
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018941575482006527,
+      "loss": 0.3371,
+      "step": 568
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018937847133790637,
+      "loss": 0.2995,
+      "step": 569
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018934112598737777,
+      "loss": 0.3038,
+      "step": 570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018930371879433025,
+      "loss": 0.3156,
+      "step": 571
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018926624978465731,
+      "loss": 0.3767,
+      "step": 572
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001892287189842952,
+      "loss": 0.2775,
+      "step": 573
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018919112641922303,
+      "loss": 0.3759,
+      "step": 574
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001891534721154626,
+      "loss": 0.3289,
+      "step": 575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018911575609907848,
+      "loss": 0.2017,
+      "step": 576
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018907797839617794,
+      "loss": 0.0915,
+      "step": 577
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000189040139032911,
+      "loss": 0.4022,
+      "step": 578
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00018900223803547023,
+      "loss": 0.4652,
+      "step": 579
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00018896427543009102,
+      "loss": 0.2639,
+      "step": 580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00018892625124305133,
+      "loss": 0.4154,
+      "step": 581
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00018888816550067175,
+      "loss": 0.3917,
+      "step": 582
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00018885001822931545,
+      "loss": 0.1619,
+      "step": 583
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00018881180945538826,
+      "loss": 0.2735,
+      "step": 584
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00018877353920533852,
+      "loss": 0.3536,
+      "step": 585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00018873520750565718,
+      "loss": 0.3287,
+      "step": 586
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00018869681438287765,
+      "loss": 0.3321,
+      "step": 587
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00018865835986357597,
+      "loss": 0.3887,
+      "step": 588
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001886198439743706,
+      "loss": 0.2031,
+      "step": 589
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001885812667419224,
+      "loss": 0.3277,
+      "step": 590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001885426281929349,
+      "loss": 0.3978,
+      "step": 591
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001885039283541539,
+      "loss": 0.3808,
+      "step": 592
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00018846516725236776,
+      "loss": 0.431,
+      "step": 593
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00018842634491440704,
+      "loss": 0.2437,
+      "step": 594
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00018838746136714494,
+      "loss": 0.364,
+      "step": 595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00018834851663749683,
+      "loss": 0.2402,
+      "step": 596
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00018830951075242056,
+      "loss": 0.3718,
+      "step": 597
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00018827044373891623,
+      "loss": 0.3139,
+      "step": 598
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00018823131562402628,
+      "loss": 0.3432,
+      "step": 599
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001881921264348355,
+      "loss": 0.3933,
+      "step": 600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00018815287619847087,
+      "loss": 0.4458,
+      "step": 601
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00018811356494210165,
+      "loss": 0.3793,
+      "step": 602
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001880741926929394,
+      "loss": 0.3546,
+      "step": 603
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00018803475947823782,
+      "loss": 0.2399,
+      "step": 604
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00018799526532529282,
+      "loss": 0.3304,
+      "step": 605
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00018795571026144255,
+      "loss": 0.2823,
+      "step": 606
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00018791609431406724,
+      "loss": 0.238,
+      "step": 607
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001878764175105894,
+      "loss": 0.242,
+      "step": 608
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00018783667987847352,
+      "loss": 0.413,
+      "step": 609
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00018779688144522622,
+      "loss": 0.3367,
+      "step": 610
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00018775702223839628,
+      "loss": 0.222,
+      "step": 611
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00018771710228557452,
+      "loss": 0.3536,
+      "step": 612
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00018767712161439374,
+      "loss": 0.2399,
+      "step": 613
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00018763708025252887,
+      "loss": 0.2757,
+      "step": 614
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00018759697822769681,
+      "loss": 0.3152,
+      "step": 615
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018755681556765645,
+      "loss": 0.3451,
+      "step": 616
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018751659230020862,
+      "loss": 0.4001,
+      "step": 617
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018747630845319612,
+      "loss": 0.3261,
+      "step": 618
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018743596405450377,
+      "loss": 0.2765,
+      "step": 619
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018739555913205815,
+      "loss": 0.3791,
+      "step": 620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018735509371382785,
+      "loss": 0.2295,
+      "step": 621
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018731456782782328,
+      "loss": 0.5221,
+      "step": 622
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018727398150209675,
+      "loss": 0.2809,
+      "step": 623
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001872333347647424,
+      "loss": 0.3125,
+      "step": 624
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018719262764389615,
+      "loss": 0.3124,
+      "step": 625
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001871518601677357,
+      "loss": 0.2489,
+      "step": 626
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001871110323644806,
+      "loss": 0.3399,
+      "step": 627
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018707014426239216,
+      "loss": 0.2163,
+      "step": 628
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018702919588977336,
+      "loss": 0.3024,
+      "step": 629
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001869881872749689,
+      "loss": 0.3693,
+      "step": 630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001869471184463653,
+      "loss": 0.4264,
+      "step": 631
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001869059894323906,
+      "loss": 0.1319,
+      "step": 632
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018686480026151457,
+      "loss": 0.2647,
+      "step": 633
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018682355096224872,
+      "loss": 0.3384,
+      "step": 634
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000186782241563146,
+      "loss": 0.3035,
+      "step": 635
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001867408720928011,
+      "loss": 0.35,
+      "step": 636
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001866994425798502,
+      "loss": 0.3296,
+      "step": 637
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001866579530529711,
+      "loss": 0.2873,
+      "step": 638
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018661640354088315,
+      "loss": 0.3217,
+      "step": 639
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018657479407234714,
+      "loss": 0.3347,
+      "step": 640
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018653312467616545,
+      "loss": 0.4092,
+      "step": 641
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018649139538118195,
+      "loss": 0.3355,
+      "step": 642
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001864496062162819,
+      "loss": 0.3617,
+      "step": 643
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018640775721039202,
+      "loss": 0.165,
+      "step": 644
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018636584839248048,
+      "loss": 0.3588,
+      "step": 645
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018632387979155687,
+      "loss": 0.2706,
+      "step": 646
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001862818514366721,
+      "loss": 0.3457,
+      "step": 647
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001862397633569185,
+      "loss": 0.1984,
+      "step": 648
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018619761558142964,
+      "loss": 0.317,
+      "step": 649
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001861554081393806,
+      "loss": 0.432,
+      "step": 650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018611314105998762,
+      "loss": 0.2937,
+      "step": 651
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018607081437250824,
+      "loss": 0.3076,
+      "step": 652
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018602842810624124,
+      "loss": 0.3591,
+      "step": 653
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001859859822905267,
+      "loss": 0.2956,
+      "step": 654
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018594347695474593,
+      "loss": 0.3581,
+      "step": 655
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018590091212832135,
+      "loss": 0.268,
+      "step": 656
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018585828784071663,
+      "loss": 0.2474,
+      "step": 657
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018581560412143661,
+      "loss": 0.3519,
+      "step": 658
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018577286100002723,
+      "loss": 0.4074,
+      "step": 659
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018573005850607555,
+      "loss": 0.3848,
+      "step": 660
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018568719666920973,
+      "loss": 0.3465,
+      "step": 661
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018564427551909902,
+      "loss": 0.3662,
+      "step": 662
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001856012950854538,
+      "loss": 0.2099,
+      "step": 663
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001855582553980253,
+      "loss": 0.3098,
+      "step": 664
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001855151564866059,
+      "loss": 0.3158,
+      "step": 665
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018547199838102904,
+      "loss": 0.2888,
+      "step": 666
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018542878111116898,
+      "loss": 0.3291,
+      "step": 667
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018538550470694097,
+      "loss": 0.3434,
+      "step": 668
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018534216919830125,
+      "loss": 0.3885,
+      "step": 669
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018529877461524697,
+      "loss": 0.263,
+      "step": 670
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018525532098781616,
+      "loss": 0.3676,
+      "step": 671
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018521180834608762,
+      "loss": 0.3249,
+      "step": 672
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001851682367201812,
+      "loss": 0.249,
+      "step": 673
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001851246061402574,
+      "loss": 0.2799,
+      "step": 674
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001850809166365176,
+      "loss": 0.2363,
+      "step": 675
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000185037168239204,
+      "loss": 0.3668,
+      "step": 676
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018499336097859947,
+      "loss": 0.2529,
+      "step": 677
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018494949488502774,
+      "loss": 0.2634,
+      "step": 678
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001849055699888532,
+      "loss": 0.2206,
+      "step": 679
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018486158632048097,
+      "loss": 0.291,
+      "step": 680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018481754391035677,
+      "loss": 0.2579,
+      "step": 681
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001847734427889671,
+      "loss": 0.3266,
+      "step": 682
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018472928298683905,
+      "loss": 0.3593,
+      "step": 683
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018468506453454033,
+      "loss": 0.2697,
+      "step": 684
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001846407874626792,
+      "loss": 0.2856,
+      "step": 685
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001845964518019046,
+      "loss": 0.2812,
+      "step": 686
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018455205758290594,
+      "loss": 0.3632,
+      "step": 687
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018450760483641317,
+      "loss": 0.3009,
+      "step": 688
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018446309359319683,
+      "loss": 0.4158,
+      "step": 689
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018441852388406785,
+      "loss": 0.3867,
+      "step": 690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018437389573987772,
+      "loss": 0.2218,
+      "step": 691
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001843292091915183,
+      "loss": 0.1111,
+      "step": 692
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018428446426992193,
+      "loss": 0.229,
+      "step": 693
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018423966100606132,
+      "loss": 0.2935,
+      "step": 694
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001841947994309496,
+      "loss": 0.2831,
+      "step": 695
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018414987957564028,
+      "loss": 0.2941,
+      "step": 696
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018410490147122713,
+      "loss": 0.227,
+      "step": 697
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018405986514884434,
+      "loss": 0.2881,
+      "step": 698
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001840147706396663,
+      "loss": 0.1922,
+      "step": 699
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018396961797490776,
+      "loss": 0.2996,
+      "step": 700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001839244071858237,
+      "loss": 0.2432,
+      "step": 701
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001838791383037093,
+      "loss": 0.4229,
+      "step": 702
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018383381135990003,
+      "loss": 0.3569,
+      "step": 703
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018378842638577144,
+      "loss": 0.3698,
+      "step": 704
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018374298341273932,
+      "loss": 0.272,
+      "step": 705
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018369748247225965,
+      "loss": 0.1821,
+      "step": 706
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001836519235958284,
+      "loss": 0.2802,
+      "step": 707
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018360630681498182,
+      "loss": 0.2001,
+      "step": 708
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018356063216129606,
+      "loss": 0.243,
+      "step": 709
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001835148996663875,
+      "loss": 0.3062,
+      "step": 710
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001834691093619124,
+      "loss": 0.2812,
+      "step": 711
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018342326127956712,
+      "loss": 0.2993,
+      "step": 712
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018337735545108808,
+      "loss": 0.1991,
+      "step": 713
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001833313919082515,
+      "loss": 0.452,
+      "step": 714
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018328537068287374,
+      "loss": 0.326,
+      "step": 715
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018323929180681092,
+      "loss": 0.4507,
+      "step": 716
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001831931553119592,
+      "loss": 0.3643,
+      "step": 717
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018314696123025454,
+      "loss": 0.1835,
+      "step": 718
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018310070959367278,
+      "loss": 0.1948,
+      "step": 719
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018305440043422962,
+      "loss": 0.1967,
+      "step": 720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001830080337839806,
+      "loss": 0.3344,
+      "step": 721
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018296160967502098,
+      "loss": 0.3147,
+      "step": 722
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018291512813948583,
+      "loss": 0.3631,
+      "step": 723
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018286858920954998,
+      "loss": 0.3138,
+      "step": 724
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018282199291742805,
+      "loss": 0.4408,
+      "step": 725
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018277533929537422,
+      "loss": 0.1843,
+      "step": 726
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018272862837568246,
+      "loss": 0.2158,
+      "step": 727
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018268186019068642,
+      "loss": 0.347,
+      "step": 728
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018263503477275927,
+      "loss": 0.2315,
+      "step": 729
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018258815215431396,
+      "loss": 0.329,
+      "step": 730
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018254121236780288,
+      "loss": 0.2324,
+      "step": 731
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001824942154457181,
+      "loss": 0.1822,
+      "step": 732
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018244716142059114,
+      "loss": 0.2761,
+      "step": 733
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001824000503249932,
+      "loss": 0.2391,
+      "step": 734
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018235288219153485,
+      "loss": 0.2526,
+      "step": 735
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018230565705286617,
+      "loss": 0.2403,
+      "step": 736
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018225837494167672,
+      "loss": 0.2416,
+      "step": 737
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018221103589069552,
+      "loss": 0.2628,
+      "step": 738
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018216363993269093,
+      "loss": 0.1825,
+      "step": 739
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018211618710047077,
+      "loss": 0.1543,
+      "step": 740
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001820686774268822,
+      "loss": 0.1856,
+      "step": 741
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018202111094481173,
+      "loss": 0.2236,
+      "step": 742
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018197348768718522,
+      "loss": 0.2864,
+      "step": 743
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018192580768696775,
+      "loss": 0.1939,
+      "step": 744
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018187807097716374,
+      "loss": 0.2532,
+      "step": 745
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001818302775908169,
+      "loss": 0.3025,
+      "step": 746
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018178242756101013,
+      "loss": 0.1924,
+      "step": 747
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001817345209208655,
+      "loss": 0.2351,
+      "step": 748
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001816865577035443,
+      "loss": 0.3442,
+      "step": 749
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018163853794224703,
+      "loss": 0.2365,
+      "step": 750
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018159046167021326,
+      "loss": 0.3609,
+      "step": 751
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001815423289207217,
+      "loss": 0.2266,
+      "step": 752
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018149413972709012,
+      "loss": 0.1991,
+      "step": 753
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001814458941226755,
+      "loss": 0.3051,
+      "step": 754
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018139759214087369,
+      "loss": 0.2871,
+      "step": 755
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018134923381511966,
+      "loss": 0.4585,
+      "step": 756
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018130081917888734,
+      "loss": 0.2535,
+      "step": 757
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018125234826568972,
+      "loss": 0.2987,
+      "step": 758
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001812038211090786,
+      "loss": 0.3086,
+      "step": 759
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018115523774264492,
+      "loss": 0.2005,
+      "step": 760
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001811065982000183,
+      "loss": 0.2692,
+      "step": 761
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001810579025148674,
+      "loss": 0.4413,
+      "step": 762
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018100915072089967,
+      "loss": 0.2543,
+      "step": 763
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001809603428518615,
+      "loss": 0.2101,
+      "step": 764
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001809114789415379,
+      "loss": 0.2243,
+      "step": 765
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018086255902375286,
+      "loss": 0.2421,
+      "step": 766
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001808135831323691,
+      "loss": 0.1715,
+      "step": 767
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000180764551301288,
+      "loss": 0.2534,
+      "step": 768
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018071546356444973,
+      "loss": 0.3424,
+      "step": 769
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018066631995583318,
+      "loss": 0.2321,
+      "step": 770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018061712050945584,
+      "loss": 0.2226,
+      "step": 771
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018056786525937394,
+      "loss": 0.2603,
+      "step": 772
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018051855423968223,
+      "loss": 0.2871,
+      "step": 773
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018046918748451418,
+      "loss": 0.3288,
+      "step": 774
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018041976502804175,
+      "loss": 0.3684,
+      "step": 775
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018037028690447547,
+      "loss": 0.4459,
+      "step": 776
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018032075314806448,
+      "loss": 0.2651,
+      "step": 777
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018027116379309638,
+      "loss": 0.204,
+      "step": 778
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001802215188738972,
+      "loss": 0.1945,
+      "step": 779
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018017181842483148,
+      "loss": 0.2372,
+      "step": 780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018012206248030224,
+      "loss": 0.2937,
+      "step": 781
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018007225107475087,
+      "loss": 0.3691,
+      "step": 782
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018002238424265712,
+      "loss": 0.2545,
+      "step": 783
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017997246201853918,
+      "loss": 0.2494,
+      "step": 784
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001799224844369535,
+      "loss": 0.1958,
+      "step": 785
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017987245153249494,
+      "loss": 0.2434,
+      "step": 786
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001798223633397966,
+      "loss": 0.1881,
+      "step": 787
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017977221989352982,
+      "loss": 0.1827,
+      "step": 788
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017972202122840426,
+      "loss": 0.2691,
+      "step": 789
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017967176737916776,
+      "loss": 0.236,
+      "step": 790
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017962145838060635,
+      "loss": 0.3316,
+      "step": 791
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017957109426754427,
+      "loss": 0.2439,
+      "step": 792
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017952067507484394,
+      "loss": 0.2275,
+      "step": 793
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017947020083740575,
+      "loss": 0.316,
+      "step": 794
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001794196715901684,
+      "loss": 0.1572,
+      "step": 795
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001793690873681085,
+      "loss": 0.3193,
+      "step": 796
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017931844820624085,
+      "loss": 0.327,
+      "step": 797
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017926775413961815,
+      "loss": 0.2915,
+      "step": 798
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001792170052033312,
+      "loss": 0.1687,
+      "step": 799
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017916620143250877,
+      "loss": 0.1305,
+      "step": 800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017911534286231754,
+      "loss": 0.2037,
+      "step": 801
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017906442952796213,
+      "loss": 0.1302,
+      "step": 802
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001790134614646851,
+      "loss": 0.3322,
+      "step": 803
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001789624387077669,
+      "loss": 0.2175,
+      "step": 804
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001789113612925258,
+      "loss": 0.2276,
+      "step": 805
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017886022925431795,
+      "loss": 0.2507,
+      "step": 806
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017880904262853727,
+      "loss": 0.1961,
+      "step": 807
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001787578014506155,
+      "loss": 0.1695,
+      "step": 808
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017870650575602207,
+      "loss": 0.2346,
+      "step": 809
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017865515558026428,
+      "loss": 0.2358,
+      "step": 810
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017860375095888698,
+      "loss": 0.1687,
+      "step": 811
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001785522919274728,
+      "loss": 0.1535,
+      "step": 812
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017850077852164207,
+      "loss": 0.237,
+      "step": 813
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001784492107770527,
+      "loss": 0.3194,
+      "step": 814
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001783975887294002,
+      "loss": 0.1756,
+      "step": 815
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001783459124144177,
+      "loss": 0.1172,
+      "step": 816
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017829418186787589,
+      "loss": 0.3665,
+      "step": 817
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017824239712558302,
+      "loss": 0.2561,
+      "step": 818
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001781905582233848,
+      "loss": 0.2071,
+      "step": 819
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017813866519716446,
+      "loss": 0.372,
+      "step": 820
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017808671808284272,
+      "loss": 0.2652,
+      "step": 821
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001780347169163777,
+      "loss": 0.2799,
+      "step": 822
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017798266173376496,
+      "loss": 0.3879,
+      "step": 823
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017793055257103743,
+      "loss": 0.2589,
+      "step": 824
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017787838946426544,
+      "loss": 0.1873,
+      "step": 825
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001778261724495566,
+      "loss": 0.2849,
+      "step": 826
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017777390156305592,
+      "loss": 0.2214,
+      "step": 827
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001777215768409456,
+      "loss": 0.3025,
+      "step": 828
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001776691983194452,
+      "loss": 0.1666,
+      "step": 829
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001776167660348115,
+      "loss": 0.2567,
+      "step": 830
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017756428002333843,
+      "loss": 0.1155,
+      "step": 831
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017751174032135716,
+      "loss": 0.214,
+      "step": 832
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017745914696523602,
+      "loss": 0.3534,
+      "step": 833
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001774064999913805,
+      "loss": 0.292,
+      "step": 834
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001773537994362332,
+      "loss": 0.3356,
+      "step": 835
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001773010453362737,
+      "loss": 0.2953,
+      "step": 836
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017724823772801885,
+      "loss": 0.1922,
+      "step": 837
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017719537664802234,
+      "loss": 0.2613,
+      "step": 838
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017714246213287505,
+      "loss": 0.1698,
+      "step": 839
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017708949421920463,
+      "loss": 0.2185,
+      "step": 840
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017703647294367597,
+      "loss": 0.1696,
+      "step": 841
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017698339834299061,
+      "loss": 0.2894,
+      "step": 842
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017693027045388726,
+      "loss": 0.2221,
+      "step": 843
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017687708931314134,
+      "loss": 0.3483,
+      "step": 844
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017682385495756514,
+      "loss": 0.1336,
+      "step": 845
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001767705674240079,
+      "loss": 0.2697,
+      "step": 846
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017671722674935557,
+      "loss": 0.2868,
+      "step": 847
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017666383297053098,
+      "loss": 0.267,
+      "step": 848
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017661038612449358,
+      "loss": 0.295,
+      "step": 849
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017655688624823967,
+      "loss": 0.3896,
+      "step": 850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001765033333788022,
+      "loss": 0.3024,
+      "step": 851
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017644972755325084,
+      "loss": 0.1977,
+      "step": 852
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017639606880869187,
+      "loss": 0.249,
+      "step": 853
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017634235718226826,
+      "loss": 0.2258,
+      "step": 854
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017628859271115953,
+      "loss": 0.2794,
+      "step": 855
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017623477543258183,
+      "loss": 0.1857,
+      "step": 856
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001761809053837878,
+      "loss": 0.2408,
+      "step": 857
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017612698260206666,
+      "loss": 0.2486,
+      "step": 858
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017607300712474416,
+      "loss": 0.1848,
+      "step": 859
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001760189789891824,
+      "loss": 0.321,
+      "step": 860
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017596489823278006,
+      "loss": 0.2427,
+      "step": 861
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001759107648929722,
+      "loss": 0.1817,
+      "step": 862
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017585657900723027,
+      "loss": 0.2924,
+      "step": 863
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017580234061306208,
+      "loss": 0.3916,
+      "step": 864
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017574804974801187,
+      "loss": 0.1743,
+      "step": 865
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017569370644966007,
+      "loss": 0.1338,
+      "step": 866
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017563931075562347,
+      "loss": 0.2866,
+      "step": 867
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001755848627035552,
+      "loss": 0.3199,
+      "step": 868
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017553036233114447,
+      "loss": 0.234,
+      "step": 869
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017547580967611684,
+      "loss": 0.2246,
+      "step": 870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017542120477623401,
+      "loss": 0.1688,
+      "step": 871
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001753665476692939,
+      "loss": 0.4161,
+      "step": 872
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001753118383931305,
+      "loss": 0.1771,
+      "step": 873
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017525707698561385,
+      "loss": 0.2802,
+      "step": 874
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017520226348465028,
+      "loss": 0.226,
+      "step": 875
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000175147397928182,
+      "loss": 0.3809,
+      "step": 876
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017509248035418732,
+      "loss": 0.2412,
+      "step": 877
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017503751080068052,
+      "loss": 0.2435,
+      "step": 878
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017498248930571193,
+      "loss": 0.1925,
+      "step": 879
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017492741590736778,
+      "loss": 0.2112,
+      "step": 880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017487229064377022,
+      "loss": 0.27,
+      "step": 881
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017481711355307736,
+      "loss": 0.3055,
+      "step": 882
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001747618846734831,
+      "loss": 0.2365,
+      "step": 883
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001747066040432173,
+      "loss": 0.316,
+      "step": 884
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001746512717005455,
+      "loss": 0.3422,
+      "step": 885
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017459588768376923,
+      "loss": 0.1506,
+      "step": 886
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017454045203122554,
+      "loss": 0.2084,
+      "step": 887
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017448496478128742,
+      "loss": 0.1769,
+      "step": 888
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001744294259723635,
+      "loss": 0.2949,
+      "step": 889
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017437383564289816,
+      "loss": 0.2592,
+      "step": 890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017431819383137133,
+      "loss": 0.3036,
+      "step": 891
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017426250057629865,
+      "loss": 0.3226,
+      "step": 892
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001742067559162314,
+      "loss": 0.1813,
+      "step": 893
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017415095988975636,
+      "loss": 0.201,
+      "step": 894
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017409511253549593,
+      "loss": 0.1749,
+      "step": 895
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000174039213892108,
+      "loss": 0.2207,
+      "step": 896
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000173983263998286,
+      "loss": 0.2913,
+      "step": 897
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001739272628927588,
+      "loss": 0.2617,
+      "step": 898
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00017387121061429077,
+      "loss": 0.2061,
+      "step": 899
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017381510720168163,
+      "loss": 0.193,
+      "step": 900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017375895269376655,
+      "loss": 0.2258,
+      "step": 901
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017370274712941606,
+      "loss": 0.2811,
+      "step": 902
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017364649054753603,
+      "loss": 0.2453,
+      "step": 903
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001735901829870676,
+      "loss": 0.1704,
+      "step": 904
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017353382448698725,
+      "loss": 0.1792,
+      "step": 905
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017347741508630672,
+      "loss": 0.2441,
+      "step": 906
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000173420954824073,
+      "loss": 0.2107,
+      "step": 907
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001733644437393682,
+      "loss": 0.1426,
+      "step": 908
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001733078818713097,
+      "loss": 0.2933,
+      "step": 909
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017325126925904998,
+      "loss": 0.2136,
+      "step": 910
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017319460594177668,
+      "loss": 0.2711,
+      "step": 911
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001731378919587125,
+      "loss": 0.1356,
+      "step": 912
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001730811273491153,
+      "loss": 0.3089,
+      "step": 913
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017302431215227782,
+      "loss": 0.2504,
+      "step": 914
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017296744640752802,
+      "loss": 0.1675,
+      "step": 915
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017291053015422867,
+      "loss": 0.1151,
+      "step": 916
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001728535634317776,
+      "loss": 0.3119,
+      "step": 917
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00017279654627960756,
+      "loss": 0.3784,
+      "step": 918
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017273947873718617,
+      "loss": 0.2807,
+      "step": 919
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000172682360844016,
+      "loss": 0.1607,
+      "step": 920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017262519263963442,
+      "loss": 0.2942,
+      "step": 921
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017256797416361362,
+      "loss": 0.2008,
+      "step": 922
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017251070545556063,
+      "loss": 0.1764,
+      "step": 923
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017245338655511724,
+      "loss": 0.1843,
+      "step": 924
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017239601750195999,
+      "loss": 0.2535,
+      "step": 925
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017233859833580003,
+      "loss": 0.2943,
+      "step": 926
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017228112909638334,
+      "loss": 0.2516,
+      "step": 927
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017222360982349053,
+      "loss": 0.2925,
+      "step": 928
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001721660405569368,
+      "loss": 0.3369,
+      "step": 929
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017210842133657197,
+      "loss": 0.2365,
+      "step": 930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017205075220228045,
+      "loss": 0.2441,
+      "step": 931
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017199303319398123,
+      "loss": 0.2055,
+      "step": 932
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017193526435162773,
+      "loss": 0.2948,
+      "step": 933
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000171877445715208,
+      "loss": 0.1295,
+      "step": 934
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001718195773247444,
+      "loss": 0.1554,
+      "step": 935
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017176165922029387,
+      "loss": 0.1863,
+      "step": 936
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001717036914419477,
+      "loss": 0.2909,
+      "step": 937
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017164567402983152,
+      "loss": 0.2992,
+      "step": 938
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017158760702410542,
+      "loss": 0.1382,
+      "step": 939
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017152949046496377,
+      "loss": 0.1389,
+      "step": 940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017147132439263518,
+      "loss": 0.2193,
+      "step": 941
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017141310884738267,
+      "loss": 0.3325,
+      "step": 942
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017135484386950335,
+      "loss": 0.2249,
+      "step": 943
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001712965294993287,
+      "loss": 0.2402,
+      "step": 944
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001712381657772242,
+      "loss": 0.2994,
+      "step": 945
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017117975274358975,
+      "loss": 0.2955,
+      "step": 946
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017112129043885915,
+      "loss": 0.1845,
+      "step": 947
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017106277890350038,
+      "loss": 0.0992,
+      "step": 948
+    },
+    {
+      "epoch": 0.5,
+      "eval_loss": 0.2248118370771408,
+      "eval_runtime": 683.3704,
+      "eval_samples_per_second": 9.815,
+      "eval_steps_per_second": 2.454,
+      "step": 948
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017100421817801552,
+      "loss": 0.2319,
+      "step": 949
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017094560830294074,
+      "loss": 0.199,
+      "step": 950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017088694931884617,
+      "loss": 0.2376,
+      "step": 951
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017082824126633591,
+      "loss": 0.2192,
+      "step": 952
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001707694841860481,
+      "loss": 0.2483,
+      "step": 953
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017071067811865476,
+      "loss": 0.3265,
+      "step": 954
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017065182310486185,
+      "loss": 0.1277,
+      "step": 955
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001705929191854092,
+      "loss": 0.1946,
+      "step": 956
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017053396640107047,
+      "loss": 0.1577,
+      "step": 957
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017047496479265318,
+      "loss": 0.238,
+      "step": 958
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017041591440099867,
+      "loss": 0.1685,
+      "step": 959
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017035681526698197,
+      "loss": 0.1262,
+      "step": 960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017029766743151189,
+      "loss": 0.2205,
+      "step": 961
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017023847093553102,
+      "loss": 0.2684,
+      "step": 962
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017017922582001545,
+      "loss": 0.2532,
+      "step": 963
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017011993212597514,
+      "loss": 0.2328,
+      "step": 964
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017006058989445352,
+      "loss": 0.2268,
+      "step": 965
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001700011991665277,
+      "loss": 0.2321,
+      "step": 966
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016994175998330836,
+      "loss": 0.1233,
+      "step": 967
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016988227238593962,
+      "loss": 0.2376,
+      "step": 968
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001698227364155993,
+      "loss": 0.2735,
+      "step": 969
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001697631521134985,
+      "loss": 0.1292,
+      "step": 970
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016970351952088188,
+      "loss": 0.1159,
+      "step": 971
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016964383867902757,
+      "loss": 0.1861,
+      "step": 972
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016958410962924697,
+      "loss": 0.1998,
+      "step": 973
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000169524332412885,
+      "loss": 0.2399,
+      "step": 974
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016946450707131978,
+      "loss": 0.2197,
+      "step": 975
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016940463364596283,
+      "loss": 0.2019,
+      "step": 976
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016934471217825898,
+      "loss": 0.2025,
+      "step": 977
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001692847427096862,
+      "loss": 0.2717,
+      "step": 978
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016922472528175576,
+      "loss": 0.1484,
+      "step": 979
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016916465993601216,
+      "loss": 0.1891,
+      "step": 980
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016910454671403299,
+      "loss": 0.318,
+      "step": 981
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016904438565742904,
+      "loss": 0.2791,
+      "step": 982
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016898417680784422,
+      "loss": 0.1819,
+      "step": 983
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016892392020695542,
+      "loss": 0.3261,
+      "step": 984
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001688636158964727,
+      "loss": 0.304,
+      "step": 985
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016880326391813916,
+      "loss": 0.1509,
+      "step": 986
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001687428643137307,
+      "loss": 0.1823,
+      "step": 987
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001686824171250565,
+      "loss": 0.2557,
+      "step": 988
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016862192239395837,
+      "loss": 0.2367,
+      "step": 989
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001685613801623112,
+      "loss": 0.1881,
+      "step": 990
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016850079047202275,
+      "loss": 0.2442,
+      "step": 991
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016844015336503357,
+      "loss": 0.1624,
+      "step": 992
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016837946888331707,
+      "loss": 0.2578,
+      "step": 993
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016831873706887952,
+      "loss": 0.1803,
+      "step": 994
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016825795796375978,
+      "loss": 0.159,
+      "step": 995
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001681971316100296,
+      "loss": 0.2785,
+      "step": 996
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016813625804979338,
+      "loss": 0.2227,
+      "step": 997
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016807533732518821,
+      "loss": 0.1191,
+      "step": 998
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016801436947838384,
+      "loss": 0.3066,
+      "step": 999
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016795335455158252,
+      "loss": 0.1774,
+      "step": 1000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001678922925870193,
+      "loss": 0.1843,
+      "step": 1001
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016783118362696163,
+      "loss": 0.1607,
+      "step": 1002
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016777002771370953,
+      "loss": 0.2425,
+      "step": 1003
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016770882488959554,
+      "loss": 0.1218,
+      "step": 1004
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016764757519698465,
+      "loss": 0.2003,
+      "step": 1005
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001675862786782743,
+      "loss": 0.1496,
+      "step": 1006
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016752493537589437,
+      "loss": 0.3887,
+      "step": 1007
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016746354533230705,
+      "loss": 0.1598,
+      "step": 1008
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000167402108590007,
+      "loss": 0.1702,
+      "step": 1009
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016734062519152112,
+      "loss": 0.2758,
+      "step": 1010
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001672790951794086,
+      "loss": 0.2565,
+      "step": 1011
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016721751859626095,
+      "loss": 0.2822,
+      "step": 1012
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016715589548470185,
+      "loss": 0.2398,
+      "step": 1013
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016709422588738726,
+      "loss": 0.2,
+      "step": 1014
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016703250984700528,
+      "loss": 0.2886,
+      "step": 1015
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016697074740627615,
+      "loss": 0.2014,
+      "step": 1016
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016690893860795227,
+      "loss": 0.2151,
+      "step": 1017
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016684708349481804,
+      "loss": 0.2235,
+      "step": 1018
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016678518210969,
+      "loss": 0.1507,
+      "step": 1019
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001667232344954168,
+      "loss": 0.281,
+      "step": 1020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016666124069487882,
+      "loss": 0.2057,
+      "step": 1021
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016659920075098866,
+      "loss": 0.2804,
+      "step": 1022
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016653711470669084,
+      "loss": 0.103,
+      "step": 1023
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016647498260496165,
+      "loss": 0.2476,
+      "step": 1024
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016641280448880934,
+      "loss": 0.2261,
+      "step": 1025
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016635058040127406,
+      "loss": 0.2407,
+      "step": 1026
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001662883103854277,
+      "loss": 0.282,
+      "step": 1027
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000166225994484374,
+      "loss": 0.2101,
+      "step": 1028
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016616363274124844,
+      "loss": 0.2921,
+      "step": 1029
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016610122519921817,
+      "loss": 0.1365,
+      "step": 1030
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001660387719014822,
+      "loss": 0.2909,
+      "step": 1031
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001659762728912711,
+      "loss": 0.2805,
+      "step": 1032
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016591372821184703,
+      "loss": 0.2616,
+      "step": 1033
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016585113790650388,
+      "loss": 0.1887,
+      "step": 1034
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016578850201856704,
+      "loss": 0.0801,
+      "step": 1035
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016572582059139353,
+      "loss": 0.2332,
+      "step": 1036
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016566309366837186,
+      "loss": 0.1621,
+      "step": 1037
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016560032129292197,
+      "loss": 0.1919,
+      "step": 1038
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016553750350849534,
+      "loss": 0.3352,
+      "step": 1039
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016547464035857482,
+      "loss": 0.2112,
+      "step": 1040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001654117318866748,
+      "loss": 0.0951,
+      "step": 1041
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001653487781363408,
+      "loss": 0.2601,
+      "step": 1042
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016528577915114992,
+      "loss": 0.3096,
+      "step": 1043
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016522273497471046,
+      "loss": 0.1938,
+      "step": 1044
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016515964565066199,
+      "loss": 0.3066,
+      "step": 1045
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001650965112226754,
+      "loss": 0.1719,
+      "step": 1046
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001650333317344527,
+      "loss": 0.1368,
+      "step": 1047
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016497010722972717,
+      "loss": 0.2769,
+      "step": 1048
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001649068377522633,
+      "loss": 0.2058,
+      "step": 1049
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016484352334585653,
+      "loss": 0.1596,
+      "step": 1050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001647801640543336,
+      "loss": 0.1181,
+      "step": 1051
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016471675992155217,
+      "loss": 0.1843,
+      "step": 1052
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016465331099140102,
+      "loss": 0.2563,
+      "step": 1053
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016458981730779998,
+      "loss": 0.1575,
+      "step": 1054
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001645262789146997,
+      "loss": 0.1867,
+      "step": 1055
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016446269585608194,
+      "loss": 0.2861,
+      "step": 1056
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016439906817595929,
+      "loss": 0.2191,
+      "step": 1057
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016433539591837525,
+      "loss": 0.1784,
+      "step": 1058
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001642716791274042,
+      "loss": 0.3081,
+      "step": 1059
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001642079178471513,
+      "loss": 0.274,
+      "step": 1060
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016414411212175255,
+      "loss": 0.1619,
+      "step": 1061
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016408026199537468,
+      "loss": 0.1497,
+      "step": 1062
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016401636751221517,
+      "loss": 0.13,
+      "step": 1063
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001639524287165022,
+      "loss": 0.2481,
+      "step": 1064
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016388844565249464,
+      "loss": 0.2375,
+      "step": 1065
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016382441836448202,
+      "loss": 0.3027,
+      "step": 1066
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001637603468967844,
+      "loss": 0.2743,
+      "step": 1067
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001636962312937525,
+      "loss": 0.1273,
+      "step": 1068
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016363207159976755,
+      "loss": 0.0966,
+      "step": 1069
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016356786785924135,
+      "loss": 0.1976,
+      "step": 1070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001635036201166161,
+      "loss": 0.2194,
+      "step": 1071
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016343932841636456,
+      "loss": 0.1342,
+      "step": 1072
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016337499280298984,
+      "loss": 0.1649,
+      "step": 1073
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001633106133210255,
+      "loss": 0.1314,
+      "step": 1074
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016324619001503545,
+      "loss": 0.1961,
+      "step": 1075
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016318172292961389,
+      "loss": 0.2579,
+      "step": 1076
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001631172121093854,
+      "loss": 0.2228,
+      "step": 1077
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016305265759900477,
+      "loss": 0.2131,
+      "step": 1078
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016298805944315708,
+      "loss": 0.2774,
+      "step": 1079
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016292341768655758,
+      "loss": 0.2261,
+      "step": 1080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016285873237395176,
+      "loss": 0.2141,
+      "step": 1081
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001627940035501152,
+      "loss": 0.1791,
+      "step": 1082
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016272923125985363,
+      "loss": 0.2808,
+      "step": 1083
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001626644155480028,
+      "loss": 0.1548,
+      "step": 1084
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016259955645942864,
+      "loss": 0.2639,
+      "step": 1085
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016253465403902702,
+      "loss": 0.1705,
+      "step": 1086
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016246970833172383,
+      "loss": 0.2487,
+      "step": 1087
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016240471938247492,
+      "loss": 0.2088,
+      "step": 1088
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016233968723626608,
+      "loss": 0.212,
+      "step": 1089
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000162274611938113,
+      "loss": 0.195,
+      "step": 1090
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016220949353306114,
+      "loss": 0.2121,
+      "step": 1091
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000162144332066186,
+      "loss": 0.2462,
+      "step": 1092
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016207912758259272,
+      "loss": 0.2408,
+      "step": 1093
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016201388012741634,
+      "loss": 0.2541,
+      "step": 1094
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001619485897458215,
+      "loss": 0.2374,
+      "step": 1095
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016188325648300268,
+      "loss": 0.231,
+      "step": 1096
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016181788038418398,
+      "loss": 0.175,
+      "step": 1097
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001617524614946192,
+      "loss": 0.1361,
+      "step": 1098
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001616869998595917,
+      "loss": 0.1256,
+      "step": 1099
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016162149552441445,
+      "loss": 0.2966,
+      "step": 1100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016155594853443,
+      "loss": 0.1651,
+      "step": 1101
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001614903589350104,
+      "loss": 0.3146,
+      "step": 1102
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016142472677155723,
+      "loss": 0.2281,
+      "step": 1103
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001613590520895015,
+      "loss": 0.1834,
+      "step": 1104
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016129333493430364,
+      "loss": 0.1806,
+      "step": 1105
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016122757535145347,
+      "loss": 0.2328,
+      "step": 1106
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001611617733864703,
+      "loss": 0.2483,
+      "step": 1107
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016109592908490261,
+      "loss": 0.2097,
+      "step": 1108
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016103004249232828,
+      "loss": 0.1533,
+      "step": 1109
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001609641136543545,
+      "loss": 0.1207,
+      "step": 1110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016089814261661752,
+      "loss": 0.1884,
+      "step": 1111
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001608321294247831,
+      "loss": 0.1436,
+      "step": 1112
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016076607412454584,
+      "loss": 0.1678,
+      "step": 1113
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001606999767616298,
+      "loss": 0.0973,
+      "step": 1114
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001606338373817879,
+      "loss": 0.2954,
+      "step": 1115
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001605676560308023,
+      "loss": 0.1868,
+      "step": 1116
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001605014327544842,
+      "loss": 0.117,
+      "step": 1117
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016043516759867374,
+      "loss": 0.1519,
+      "step": 1118
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016036886060924013,
+      "loss": 0.2882,
+      "step": 1119
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001603025118320815,
+      "loss": 0.2382,
+      "step": 1120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001602361213131249,
+      "loss": 0.107,
+      "step": 1121
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016016968909832632,
+      "loss": 0.1723,
+      "step": 1122
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016010321523367055,
+      "loss": 0.2484,
+      "step": 1123
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016003669976517125,
+      "loss": 0.2224,
+      "step": 1124
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015997014273887085,
+      "loss": 0.3619,
+      "step": 1125
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015990354420084056,
+      "loss": 0.1407,
+      "step": 1126
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015983690419718038,
+      "loss": 0.248,
+      "step": 1127
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015977022277401893,
+      "loss": 0.1617,
+      "step": 1128
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015970349997751355,
+      "loss": 0.1832,
+      "step": 1129
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015963673585385016,
+      "loss": 0.1627,
+      "step": 1130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015956993044924334,
+      "loss": 0.2363,
+      "step": 1131
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015950308380993625,
+      "loss": 0.1417,
+      "step": 1132
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001594361959822006,
+      "loss": 0.2745,
+      "step": 1133
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015936926701233653,
+      "loss": 0.289,
+      "step": 1134
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015930229694667276,
+      "loss": 0.1142,
+      "step": 1135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001592352858315664,
+      "loss": 0.2093,
+      "step": 1136
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015916823371340296,
+      "loss": 0.1656,
+      "step": 1137
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001591011406385964,
+      "loss": 0.1432,
+      "step": 1138
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000159034006653589,
+      "loss": 0.2203,
+      "step": 1139
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001589668318048513,
+      "loss": 0.1496,
+      "step": 1140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001588996161388822,
+      "loss": 0.2327,
+      "step": 1141
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015883235970220886,
+      "loss": 0.1868,
+      "step": 1142
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015876506254138657,
+      "loss": 0.1784,
+      "step": 1143
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001586977247029989,
+      "loss": 0.1543,
+      "step": 1144
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015863034623365754,
+      "loss": 0.2577,
+      "step": 1145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015856292718000235,
+      "loss": 0.1713,
+      "step": 1146
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001584954675887012,
+      "loss": 0.1933,
+      "step": 1147
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015842796750645008,
+      "loss": 0.251,
+      "step": 1148
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000158360426979973,
+      "loss": 0.2689,
+      "step": 1149
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015829284605602193,
+      "loss": 0.1312,
+      "step": 1150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015822522478137693,
+      "loss": 0.1173,
+      "step": 1151
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015815756320284577,
+      "loss": 0.1227,
+      "step": 1152
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015808986136726428,
+      "loss": 0.2088,
+      "step": 1153
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015802211932149614,
+      "loss": 0.2212,
+      "step": 1154
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015795433711243286,
+      "loss": 0.2685,
+      "step": 1155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015788651478699367,
+      "loss": 0.1998,
+      "step": 1156
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015781865239212573,
+      "loss": 0.274,
+      "step": 1157
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015775074997480374,
+      "loss": 0.1282,
+      "step": 1158
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001576828075820303,
+      "loss": 0.3618,
+      "step": 1159
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015761482526083546,
+      "loss": 0.2845,
+      "step": 1160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015754680305827713,
+      "loss": 0.1266,
+      "step": 1161
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001574787410214407,
+      "loss": 0.1431,
+      "step": 1162
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015741063919743917,
+      "loss": 0.1556,
+      "step": 1163
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015734249763341306,
+      "loss": 0.2614,
+      "step": 1164
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015727431637653041,
+      "loss": 0.238,
+      "step": 1165
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015720609547398677,
+      "loss": 0.2459,
+      "step": 1166
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015713783497300505,
+      "loss": 0.1817,
+      "step": 1167
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015706953492083568,
+      "loss": 0.0959,
+      "step": 1168
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015700119536475634,
+      "loss": 0.1364,
+      "step": 1169
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015693281635207216,
+      "loss": 0.117,
+      "step": 1170
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015686439793011556,
+      "loss": 0.2701,
+      "step": 1171
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015679594014624622,
+      "loss": 0.207,
+      "step": 1172
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015672744304785104,
+      "loss": 0.1699,
+      "step": 1173
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001566589066823442,
+      "loss": 0.2169,
+      "step": 1174
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015659033109716698,
+      "loss": 0.0678,
+      "step": 1175
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015652171633978793,
+      "loss": 0.2328,
+      "step": 1176
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015645306245770256,
+      "loss": 0.2603,
+      "step": 1177
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001563843694984336,
+      "loss": 0.2114,
+      "step": 1178
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015631563750953076,
+      "loss": 0.2167,
+      "step": 1179
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015624686653857074,
+      "loss": 0.3016,
+      "step": 1180
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001561780566331573,
+      "loss": 0.1622,
+      "step": 1181
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015610920784092107,
+      "loss": 0.1814,
+      "step": 1182
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015604032020951968,
+      "loss": 0.182,
+      "step": 1183
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001559713937866376,
+      "loss": 0.168,
+      "step": 1184
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015590242861998616,
+      "loss": 0.1222,
+      "step": 1185
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001558334247573035,
+      "loss": 0.1175,
+      "step": 1186
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001557643822463545,
+      "loss": 0.2246,
+      "step": 1187
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015569530113493095,
+      "loss": 0.1939,
+      "step": 1188
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015562618147085114,
+      "loss": 0.2485,
+      "step": 1189
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015555702330196023,
+      "loss": 0.333,
+      "step": 1190
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015548782667612995,
+      "loss": 0.1979,
+      "step": 1191
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015541859164125863,
+      "loss": 0.1355,
+      "step": 1192
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015534931824527127,
+      "loss": 0.18,
+      "step": 1193
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015528000653611935,
+      "loss": 0.2431,
+      "step": 1194
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015521065656178088,
+      "loss": 0.1738,
+      "step": 1195
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001551412683702604,
+      "loss": 0.3208,
+      "step": 1196
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015507184200958885,
+      "loss": 0.2392,
+      "step": 1197
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015500237752782362,
+      "loss": 0.2441,
+      "step": 1198
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015493287497304848,
+      "loss": 0.2075,
+      "step": 1199
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015486333439337357,
+      "loss": 0.0869,
+      "step": 1200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015479375583693535,
+      "loss": 0.1392,
+      "step": 1201
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015472413935189655,
+      "loss": 0.2814,
+      "step": 1202
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015465448498644616,
+      "loss": 0.2047,
+      "step": 1203
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015458479278879938,
+      "loss": 0.1823,
+      "step": 1204
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015451506280719763,
+      "loss": 0.1552,
+      "step": 1205
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015444529508990844,
+      "loss": 0.2431,
+      "step": 1206
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001543754896852255,
+      "loss": 0.1104,
+      "step": 1207
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001543056466414686,
+      "loss": 0.2062,
+      "step": 1208
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015423576600698353,
+      "loss": 0.1686,
+      "step": 1209
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001541658478301421,
+      "loss": 0.2138,
+      "step": 1210
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015409589215934218,
+      "loss": 0.2286,
+      "step": 1211
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015402589904300754,
+      "loss": 0.1782,
+      "step": 1212
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015395586852958788,
+      "loss": 0.2035,
+      "step": 1213
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015388580066755876,
+      "loss": 0.1019,
+      "step": 1214
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015381569550542166,
+      "loss": 0.2119,
+      "step": 1215
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001537455530917038,
+      "loss": 0.1624,
+      "step": 1216
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015367537347495822,
+      "loss": 0.089,
+      "step": 1217
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015360515670376375,
+      "loss": 0.1451,
+      "step": 1218
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015353490282672488,
+      "loss": 0.0915,
+      "step": 1219
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001534646118924718,
+      "loss": 0.2215,
+      "step": 1220
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001533942839496604,
+      "loss": 0.171,
+      "step": 1221
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015332391904697212,
+      "loss": 0.2381,
+      "step": 1222
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000153253517233114,
+      "loss": 0.268,
+      "step": 1223
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015318307855681867,
+      "loss": 0.2391,
+      "step": 1224
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015311260306684422,
+      "loss": 0.2322,
+      "step": 1225
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015304209081197425,
+      "loss": 0.2244,
+      "step": 1226
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015297154184101786,
+      "loss": 0.1248,
+      "step": 1227
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015290095620280945,
+      "loss": 0.2095,
+      "step": 1228
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001528303339462089,
+      "loss": 0.1651,
+      "step": 1229
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001527596751201014,
+      "loss": 0.0803,
+      "step": 1230
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015268897977339748,
+      "loss": 0.212,
+      "step": 1231
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015261824795503287,
+      "loss": 0.1429,
+      "step": 1232
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015254747971396866,
+      "loss": 0.144,
+      "step": 1233
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015247667509919104,
+      "loss": 0.3074,
+      "step": 1234
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015240583415971148,
+      "loss": 0.1789,
+      "step": 1235
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015233495694456654,
+      "loss": 0.1965,
+      "step": 1236
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015226404350281788,
+      "loss": 0.1805,
+      "step": 1237
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015219309388355226,
+      "loss": 0.1658,
+      "step": 1238
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001521221081358815,
+      "loss": 0.2434,
+      "step": 1239
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015205108630894239,
+      "loss": 0.2157,
+      "step": 1240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015198002845189665,
+      "loss": 0.1305,
+      "step": 1241
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015190893461393108,
+      "loss": 0.2222,
+      "step": 1242
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015183780484425726,
+      "loss": 0.2424,
+      "step": 1243
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015176663919211177,
+      "loss": 0.0928,
+      "step": 1244
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015169543770675577,
+      "loss": 0.1543,
+      "step": 1245
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001516242004374756,
+      "loss": 0.0936,
+      "step": 1246
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015155292743358208,
+      "loss": 0.1779,
+      "step": 1247
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001514816187444108,
+      "loss": 0.1012,
+      "step": 1248
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015141027441932216,
+      "loss": 0.2297,
+      "step": 1249
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015133889450770122,
+      "loss": 0.2623,
+      "step": 1250
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015126747905895758,
+      "loss": 0.2347,
+      "step": 1251
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015119602812252547,
+      "loss": 0.1429,
+      "step": 1252
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015112454174786373,
+      "loss": 0.3506,
+      "step": 1253
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015105301998445566,
+      "loss": 0.214,
+      "step": 1254
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015098146288180915,
+      "loss": 0.0679,
+      "step": 1255
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015090987048945652,
+      "loss": 0.1214,
+      "step": 1256
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015083824285695442,
+      "loss": 0.2185,
+      "step": 1257
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000150766580033884,
+      "loss": 0.2326,
+      "step": 1258
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015069488206985077,
+      "loss": 0.1628,
+      "step": 1259
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015062314901448452,
+      "loss": 0.0787,
+      "step": 1260
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015055138091743934,
+      "loss": 0.1122,
+      "step": 1261
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015047957782839358,
+      "loss": 0.254,
+      "step": 1262
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015040773979704982,
+      "loss": 0.1228,
+      "step": 1263
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001503358668731348,
+      "loss": 0.1028,
+      "step": 1264
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015026395910639948,
+      "loss": 0.1406,
+      "step": 1265
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015019201654661884,
+      "loss": 0.1823,
+      "step": 1266
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015012003924359202,
+      "loss": 0.212,
+      "step": 1267
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015004802724714224,
+      "loss": 0.1563,
+      "step": 1268
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014997598060711658,
+      "loss": 0.0587,
+      "step": 1269
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014990389937338624,
+      "loss": 0.249,
+      "step": 1270
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014983178359584634,
+      "loss": 0.2052,
+      "step": 1271
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014975963332441593,
+      "loss": 0.1589,
+      "step": 1272
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014968744860903782,
+      "loss": 0.089,
+      "step": 1273
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014961522949967886,
+      "loss": 0.1644,
+      "step": 1274
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014954297604632947,
+      "loss": 0.0875,
+      "step": 1275
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014947068829900405,
+      "loss": 0.1373,
+      "step": 1276
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014939836630774067,
+      "loss": 0.1384,
+      "step": 1277
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00014932601012260105,
+      "loss": 0.1602,
+      "step": 1278
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001492536197936706,
+      "loss": 0.162,
+      "step": 1279
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014918119537105844,
+      "loss": 0.1146,
+      "step": 1280
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014910873690489725,
+      "loss": 0.1141,
+      "step": 1281
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014903624444534316,
+      "loss": 0.1685,
+      "step": 1282
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000148963718042576,
+      "loss": 0.0956,
+      "step": 1283
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014889115774679906,
+      "loss": 0.305,
+      "step": 1284
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000148818563608239,
+      "loss": 0.1632,
+      "step": 1285
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014874593567714594,
+      "loss": 0.1049,
+      "step": 1286
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014867327400379347,
+      "loss": 0.1546,
+      "step": 1287
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001486005786384785,
+      "loss": 0.1571,
+      "step": 1288
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014852784963152117,
+      "loss": 0.2059,
+      "step": 1289
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014845508703326504,
+      "loss": 0.1959,
+      "step": 1290
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014838229089407683,
+      "loss": 0.1803,
+      "step": 1291
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001483094612643465,
+      "loss": 0.318,
+      "step": 1292
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014823659819448724,
+      "loss": 0.1568,
+      "step": 1293
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014816370173493533,
+      "loss": 0.1859,
+      "step": 1294
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014809077193615023,
+      "loss": 0.2267,
+      "step": 1295
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00014801780884861432,
+      "loss": 0.3156,
+      "step": 1296
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001479448125228332,
+      "loss": 0.1866,
+      "step": 1297
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00014787178300933542,
+      "loss": 0.2256,
+      "step": 1298
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00014779872035867252,
+      "loss": 0.3029,
+      "step": 1299
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00014772562462141882,
+      "loss": 0.1458,
+      "step": 1300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00014765249584817184,
+      "loss": 0.186,
+      "step": 1301
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00014757933408955168,
+      "loss": 0.3387,
+      "step": 1302
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00014750613939620144,
+      "loss": 0.1855,
+      "step": 1303
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00014743291181878697,
+      "loss": 0.3866,
+      "step": 1304
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001473596514079969,
+      "loss": 0.2217,
+      "step": 1305
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00014728635821454255,
+      "loss": 0.2811,
+      "step": 1306
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00014721303228915793,
+      "loss": 0.1642,
+      "step": 1307
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001471396736825998,
+      "loss": 0.1376,
+      "step": 1308
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001470662824456474,
+      "loss": 0.1926,
+      "step": 1309
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00014699285862910264,
+      "loss": 0.2456,
+      "step": 1310
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00014691940228379004,
+      "loss": 0.159,
+      "step": 1311
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00014684591346055648,
+      "loss": 0.1393,
+      "step": 1312
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00014677239221027149,
+      "loss": 0.2164,
+      "step": 1313
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001466988385838269,
+      "loss": 0.1871,
+      "step": 1314
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00014662525263213708,
+      "loss": 0.2302,
+      "step": 1315
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00014655163440613865,
+      "loss": 0.1361,
+      "step": 1316
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00014647798395679067,
+      "loss": 0.2063,
+      "step": 1317
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00014640430133507445,
+      "loss": 0.1093,
+      "step": 1318
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00014633058659199364,
+      "loss": 0.1759,
+      "step": 1319
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00014625683977857398,
+      "loss": 0.195,
+      "step": 1320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00014618306094586357,
+      "loss": 0.2586,
+      "step": 1321
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001461092501449326,
+      "loss": 0.1396,
+      "step": 1322
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001460354074268734,
+      "loss": 0.1524,
+      "step": 1323
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00014596153284280035,
+      "loss": 0.2524,
+      "step": 1324
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00014588762644384993,
+      "loss": 0.1327,
+      "step": 1325
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001458136882811807,
+      "loss": 0.2365,
+      "step": 1326
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00014573971840597312,
+      "loss": 0.1512,
+      "step": 1327
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00014566571686942955,
+      "loss": 0.2322,
+      "step": 1328
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00014559168372277448,
+      "loss": 0.1987,
+      "step": 1329
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00014551761901725402,
+      "loss": 0.1151,
+      "step": 1330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00014544352280413628,
+      "loss": 0.1629,
+      "step": 1331
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00014536939513471123,
+      "loss": 0.2446,
+      "step": 1332
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00014529523606029044,
+      "loss": 0.3006,
+      "step": 1333
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00014522104563220736,
+      "loss": 0.276,
+      "step": 1334
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014514682390181707,
+      "loss": 0.2363,
+      "step": 1335
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014507257092049637,
+      "loss": 0.1,
+      "step": 1336
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014499828673964364,
+      "loss": 0.1606,
+      "step": 1337
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014492397141067887,
+      "loss": 0.1774,
+      "step": 1338
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014484962498504368,
+      "loss": 0.1974,
+      "step": 1339
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014477524751420112,
+      "loss": 0.0979,
+      "step": 1340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014470083904963573,
+      "loss": 0.2117,
+      "step": 1341
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001446263996428536,
+      "loss": 0.1755,
+      "step": 1342
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014455192934538218,
+      "loss": 0.1637,
+      "step": 1343
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001444774282087703,
+      "loss": 0.1108,
+      "step": 1344
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014440289628458807,
+      "loss": 0.1364,
+      "step": 1345
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001443283336244271,
+      "loss": 0.2803,
+      "step": 1346
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014425374027990005,
+      "loss": 0.073,
+      "step": 1347
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014417911630264097,
+      "loss": 0.1393,
+      "step": 1348
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014410446174430508,
+      "loss": 0.1491,
+      "step": 1349
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014402977665656876,
+      "loss": 0.22,
+      "step": 1350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001439550610911295,
+      "loss": 0.2035,
+      "step": 1351
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014388031509970591,
+      "loss": 0.3069,
+      "step": 1352
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014380553873403774,
+      "loss": 0.2131,
+      "step": 1353
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014373073204588556,
+      "loss": 0.1682,
+      "step": 1354
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001436558950870311,
+      "loss": 0.2059,
+      "step": 1355
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014358102790927704,
+      "loss": 0.092,
+      "step": 1356
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014350613056444693,
+      "loss": 0.248,
+      "step": 1357
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014343120310438513,
+      "loss": 0.1759,
+      "step": 1358
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014335624558095702,
+      "loss": 0.1482,
+      "step": 1359
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014328125804604863,
+      "loss": 0.0913,
+      "step": 1360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001432062405515669,
+      "loss": 0.14,
+      "step": 1361
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014313119314943932,
+      "loss": 0.1466,
+      "step": 1362
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001430561158916143,
+      "loss": 0.219,
+      "step": 1363
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001429810088300608,
+      "loss": 0.1564,
+      "step": 1364
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014290587201676837,
+      "loss": 0.2567,
+      "step": 1365
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014283070550374726,
+      "loss": 0.1253,
+      "step": 1366
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014275550934302823,
+      "loss": 0.097,
+      "step": 1367
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014268028358666252,
+      "loss": 0.1767,
+      "step": 1368
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014260502828672192,
+      "loss": 0.0934,
+      "step": 1369
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001425297434952987,
+      "loss": 0.1582,
+      "step": 1370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014245442926450537,
+      "loss": 0.2114,
+      "step": 1371
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014237908564647503,
+      "loss": 0.1793,
+      "step": 1372
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014230371269336104,
+      "loss": 0.1359,
+      "step": 1373
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014222831045733702,
+      "loss": 0.2619,
+      "step": 1374
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014215287899059688,
+      "loss": 0.2032,
+      "step": 1375
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001420774183453548,
+      "loss": 0.1677,
+      "step": 1376
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001420019285738452,
+      "loss": 0.1659,
+      "step": 1377
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001419264097283225,
+      "loss": 0.189,
+      "step": 1378
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001418508618610614,
+      "loss": 0.1473,
+      "step": 1379
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001417752850243566,
+      "loss": 0.1423,
+      "step": 1380
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014169967927052293,
+      "loss": 0.1249,
+      "step": 1381
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014162404465189516,
+      "loss": 0.2195,
+      "step": 1382
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014154838122082807,
+      "loss": 0.3112,
+      "step": 1383
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014147268902969643,
+      "loss": 0.1084,
+      "step": 1384
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014139696813089485,
+      "loss": 0.132,
+      "step": 1385
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014132121857683783,
+      "loss": 0.2931,
+      "step": 1386
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014124544041995972,
+      "loss": 0.2135,
+      "step": 1387
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014116963371271464,
+      "loss": 0.2767,
+      "step": 1388
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014109379850757656,
+      "loss": 0.1927,
+      "step": 1389
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014101793485703904,
+      "loss": 0.11,
+      "step": 1390
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014094204281361544,
+      "loss": 0.2897,
+      "step": 1391
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014086612242983872,
+      "loss": 0.1441,
+      "step": 1392
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014079017375826148,
+      "loss": 0.1932,
+      "step": 1393
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014071419685145585,
+      "loss": 0.0787,
+      "step": 1394
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014063819176201366,
+      "loss": 0.2758,
+      "step": 1395
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000140562158542546,
+      "loss": 0.2632,
+      "step": 1396
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014048609724568367,
+      "loss": 0.1423,
+      "step": 1397
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014041000792407675,
+      "loss": 0.1784,
+      "step": 1398
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001403338906303948,
+      "loss": 0.1776,
+      "step": 1399
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014025774541732666,
+      "loss": 0.0947,
+      "step": 1400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014018157233758064,
+      "loss": 0.1405,
+      "step": 1401
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014010537144388416,
+      "loss": 0.0845,
+      "step": 1402
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014002914278898407,
+      "loss": 0.1475,
+      "step": 1403
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001399528864256463,
+      "loss": 0.1396,
+      "step": 1404
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013987660240665599,
+      "loss": 0.168,
+      "step": 1405
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013980029078481746,
+      "loss": 0.1548,
+      "step": 1406
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013972395161295414,
+      "loss": 0.1024,
+      "step": 1407
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013964758494390854,
+      "loss": 0.1519,
+      "step": 1408
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00013957119083054205,
+      "loss": 0.1565,
+      "step": 1409
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001394947693257353,
+      "loss": 0.1061,
+      "step": 1410
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013941832048238777,
+      "loss": 0.0608,
+      "step": 1411
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013934184435341777,
+      "loss": 0.2847,
+      "step": 1412
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001392653409917626,
+      "loss": 0.2893,
+      "step": 1413
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013918881045037846,
+      "loss": 0.1615,
+      "step": 1414
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013911225278224023,
+      "loss": 0.1766,
+      "step": 1415
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013903566804034166,
+      "loss": 0.1391,
+      "step": 1416
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013895905627769523,
+      "loss": 0.2277,
+      "step": 1417
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013888241754733208,
+      "loss": 0.1513,
+      "step": 1418
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013880575190230204,
+      "loss": 0.1631,
+      "step": 1419
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013872905939567363,
+      "loss": 0.1735,
+      "step": 1420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013865234008053387,
+      "loss": 0.1803,
+      "step": 1421
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013857559400998838,
+      "loss": 0.1897,
+      "step": 1422
+    },
+    {
+      "epoch": 0.75,
+      "eval_loss": 0.16311943531036377,
+      "eval_runtime": 683.209,
+      "eval_samples_per_second": 9.817,
+      "eval_steps_per_second": 2.455,
+      "step": 1422
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001384988212371613,
+      "loss": 0.1545,
+      "step": 1423
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001384220218151953,
+      "loss": 0.1964,
+      "step": 1424
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013834519579725133,
+      "loss": 0.1159,
+      "step": 1425
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000138268343236509,
+      "loss": 0.1972,
+      "step": 1426
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013819146418616606,
+      "loss": 0.2127,
+      "step": 1427
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013811455869943876,
+      "loss": 0.1761,
+      "step": 1428
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001380376268295616,
+      "loss": 0.1634,
+      "step": 1429
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013796066862978724,
+      "loss": 0.1931,
+      "step": 1430
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013788368415338677,
+      "loss": 0.0715,
+      "step": 1431
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013780667345364927,
+      "loss": 0.1371,
+      "step": 1432
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001377296365838821,
+      "loss": 0.1818,
+      "step": 1433
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013765257359741063,
+      "loss": 0.323,
+      "step": 1434
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013757548454757846,
+      "loss": 0.1194,
+      "step": 1435
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013749836948774707,
+      "loss": 0.1639,
+      "step": 1436
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001374212284712961,
+      "loss": 0.109,
+      "step": 1437
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013734406155162292,
+      "loss": 0.2091,
+      "step": 1438
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013726686878214313,
+      "loss": 0.1529,
+      "step": 1439
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013718965021629003,
+      "loss": 0.1716,
+      "step": 1440
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001371124059075148,
+      "loss": 0.1304,
+      "step": 1441
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013703513590928646,
+      "loss": 0.2382,
+      "step": 1442
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013695784027509185,
+      "loss": 0.1077,
+      "step": 1443
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013688051905843548,
+      "loss": 0.1538,
+      "step": 1444
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013680317231283965,
+      "loss": 0.208,
+      "step": 1445
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001367258000918443,
+      "loss": 0.1104,
+      "step": 1446
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013664840244900688,
+      "loss": 0.2242,
+      "step": 1447
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001365709794379027,
+      "loss": 0.1211,
+      "step": 1448
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001364935311121244,
+      "loss": 0.0726,
+      "step": 1449
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013641605752528224,
+      "loss": 0.1732,
+      "step": 1450
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001363385587310039,
+      "loss": 0.1335,
+      "step": 1451
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013626103478293469,
+      "loss": 0.1469,
+      "step": 1452
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013618348573473705,
+      "loss": 0.2504,
+      "step": 1453
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013610591164009103,
+      "loss": 0.0891,
+      "step": 1454
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001360283125526939,
+      "loss": 0.1633,
+      "step": 1455
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013595068852626028,
+      "loss": 0.1743,
+      "step": 1456
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013587303961452204,
+      "loss": 0.2268,
+      "step": 1457
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013579536587122828,
+      "loss": 0.1837,
+      "step": 1458
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013571766735014524,
+      "loss": 0.1251,
+      "step": 1459
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013563994410505635,
+      "loss": 0.2054,
+      "step": 1460
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013556219618976218,
+      "loss": 0.1854,
+      "step": 1461
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013548442365808036,
+      "loss": 0.1198,
+      "step": 1462
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013540662656384556,
+      "loss": 0.2281,
+      "step": 1463
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001353288049609094,
+      "loss": 0.2934,
+      "step": 1464
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001352509589031405,
+      "loss": 0.2515,
+      "step": 1465
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001351730884444245,
+      "loss": 0.2408,
+      "step": 1466
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013509519363866379,
+      "loss": 0.1965,
+      "step": 1467
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013501727453977763,
+      "loss": 0.1569,
+      "step": 1468
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013493933120170223,
+      "loss": 0.2497,
+      "step": 1469
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001348613636783904,
+      "loss": 0.1736,
+      "step": 1470
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013478337202381185,
+      "loss": 0.1352,
+      "step": 1471
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013470535629195286,
+      "loss": 0.1953,
+      "step": 1472
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013462731653681644,
+      "loss": 0.224,
+      "step": 1473
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013454925281242226,
+      "loss": 0.196,
+      "step": 1474
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013447116517280652,
+      "loss": 0.1986,
+      "step": 1475
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013439305367202205,
+      "loss": 0.1392,
+      "step": 1476
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001343149183641381,
+      "loss": 0.082,
+      "step": 1477
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001342367593032404,
+      "loss": 0.1167,
+      "step": 1478
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013415857654343127,
+      "loss": 0.1956,
+      "step": 1479
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013408037013882927,
+      "loss": 0.1579,
+      "step": 1480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013400214014356942,
+      "loss": 0.1811,
+      "step": 1481
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013392388661180303,
+      "loss": 0.0997,
+      "step": 1482
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013384560959769774,
+      "loss": 0.0995,
+      "step": 1483
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013376730915543736,
+      "loss": 0.1569,
+      "step": 1484
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000133688985339222,
+      "loss": 0.141,
+      "step": 1485
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013361063820326797,
+      "loss": 0.2677,
+      "step": 1486
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013353226780180762,
+      "loss": 0.1245,
+      "step": 1487
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013345387418908946,
+      "loss": 0.2866,
+      "step": 1488
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001333754574193781,
+      "loss": 0.1417,
+      "step": 1489
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013329701754695412,
+      "loss": 0.3077,
+      "step": 1490
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013321855462611416,
+      "loss": 0.2078,
+      "step": 1491
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001331400687111707,
+      "loss": 0.1619,
+      "step": 1492
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013306155985645231,
+      "loss": 0.1411,
+      "step": 1493
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013298302811630324,
+      "loss": 0.071,
+      "step": 1494
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013290447354508372,
+      "loss": 0.3154,
+      "step": 1495
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001328258961971698,
+      "loss": 0.0936,
+      "step": 1496
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013274729612695315,
+      "loss": 0.1365,
+      "step": 1497
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001326686733888413,
+      "loss": 0.1128,
+      "step": 1498
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013259002803725744,
+      "loss": 0.1136,
+      "step": 1499
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001325113601266404,
+      "loss": 0.1207,
+      "step": 1500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001324326697114446,
+      "loss": 0.1977,
+      "step": 1501
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001323539568461401,
+      "loss": 0.115,
+      "step": 1502
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013227522158521245,
+      "loss": 0.1634,
+      "step": 1503
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013219646398316272,
+      "loss": 0.1757,
+      "step": 1504
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013211768409450742,
+      "loss": 0.1698,
+      "step": 1505
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013203888197377855,
+      "loss": 0.148,
+      "step": 1506
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013196005767552346,
+      "loss": 0.1331,
+      "step": 1507
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013188121125430483,
+      "loss": 0.1436,
+      "step": 1508
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001318023427647007,
+      "loss": 0.1761,
+      "step": 1509
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013172345226130432,
+      "loss": 0.1878,
+      "step": 1510
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013164453979872424,
+      "loss": 0.248,
+      "step": 1511
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013156560543158417,
+      "loss": 0.1853,
+      "step": 1512
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013148664921452306,
+      "loss": 0.0739,
+      "step": 1513
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001314076712021949,
+      "loss": 0.0854,
+      "step": 1514
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013132867144926874,
+      "loss": 0.1526,
+      "step": 1515
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013124965001042879,
+      "loss": 0.3326,
+      "step": 1516
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001311706069403742,
+      "loss": 0.156,
+      "step": 1517
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001310915422938191,
+      "loss": 0.2221,
+      "step": 1518
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013101245612549256,
+      "loss": 0.1303,
+      "step": 1519
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013093334849013856,
+      "loss": 0.1734,
+      "step": 1520
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013085421944251593,
+      "loss": 0.1382,
+      "step": 1521
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001307750690373983,
+      "loss": 0.1389,
+      "step": 1522
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013069589732957412,
+      "loss": 0.2068,
+      "step": 1523
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013061670437384656,
+      "loss": 0.2016,
+      "step": 1524
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013053749022503354,
+      "loss": 0.1887,
+      "step": 1525
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001304582549379676,
+      "loss": 0.1931,
+      "step": 1526
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001303789985674959,
+      "loss": 0.1007,
+      "step": 1527
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013029972116848025,
+      "loss": 0.2514,
+      "step": 1528
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013022042279579697,
+      "loss": 0.2173,
+      "step": 1529
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000130141103504337,
+      "loss": 0.2476,
+      "step": 1530
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013006176334900556,
+      "loss": 0.1618,
+      "step": 1531
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012998240238472252,
+      "loss": 0.154,
+      "step": 1532
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000129903020666422,
+      "loss": 0.2169,
+      "step": 1533
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012982361824905261,
+      "loss": 0.1985,
+      "step": 1534
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012974419518757722,
+      "loss": 0.271,
+      "step": 1535
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012966475153697296,
+      "loss": 0.1763,
+      "step": 1536
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001295852873522313,
+      "loss": 0.1763,
+      "step": 1537
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012950580268835784,
+      "loss": 0.1879,
+      "step": 1538
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012942629760037238,
+      "loss": 0.2579,
+      "step": 1539
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012934677214330886,
+      "loss": 0.1994,
+      "step": 1540
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001292672263722154,
+      "loss": 0.1505,
+      "step": 1541
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012918766034215407,
+      "loss": 0.1558,
+      "step": 1542
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012910807410820095,
+      "loss": 0.2174,
+      "step": 1543
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012902846772544624,
+      "loss": 0.1237,
+      "step": 1544
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012894884124899398,
+      "loss": 0.1455,
+      "step": 1545
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001288691947339621,
+      "loss": 0.1516,
+      "step": 1546
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001287895282354825,
+      "loss": 0.1453,
+      "step": 1547
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012870984180870086,
+      "loss": 0.2567,
+      "step": 1548
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012863013550877658,
+      "loss": 0.1135,
+      "step": 1549
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012855040939088296,
+      "loss": 0.1534,
+      "step": 1550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012847066351020695,
+      "loss": 0.1167,
+      "step": 1551
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012839089792194912,
+      "loss": 0.1478,
+      "step": 1552
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012831111268132378,
+      "loss": 0.092,
+      "step": 1553
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001282313078435588,
+      "loss": 0.0851,
+      "step": 1554
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012815148346389564,
+      "loss": 0.1036,
+      "step": 1555
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012807163959758917,
+      "loss": 0.1571,
+      "step": 1556
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012799177629990792,
+      "loss": 0.2327,
+      "step": 1557
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001279118936261338,
+      "loss": 0.1845,
+      "step": 1558
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001278319916315621,
+      "loss": 0.1217,
+      "step": 1559
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001277520703715015,
+      "loss": 0.1934,
+      "step": 1560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012767212990127404,
+      "loss": 0.1503,
+      "step": 1561
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012759217027621505,
+      "loss": 0.1011,
+      "step": 1562
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012751219155167313,
+      "loss": 0.207,
+      "step": 1563
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012743219378301003,
+      "loss": 0.1488,
+      "step": 1564
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012735217702560076,
+      "loss": 0.1159,
+      "step": 1565
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012727214133483346,
+      "loss": 0.1978,
+      "step": 1566
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012719208676610934,
+      "loss": 0.198,
+      "step": 1567
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012711201337484276,
+      "loss": 0.0646,
+      "step": 1568
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000127031921216461,
+      "loss": 0.1694,
+      "step": 1569
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012695181034640436,
+      "loss": 0.0934,
+      "step": 1570
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012687168082012616,
+      "loss": 0.2773,
+      "step": 1571
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001267915326930926,
+      "loss": 0.0519,
+      "step": 1572
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001267113660207827,
+      "loss": 0.1498,
+      "step": 1573
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012663118085868836,
+      "loss": 0.2086,
+      "step": 1574
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012655097726231435,
+      "loss": 0.0917,
+      "step": 1575
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001264707552871781,
+      "loss": 0.1225,
+      "step": 1576
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001263905149888097,
+      "loss": 0.1194,
+      "step": 1577
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012631025642275212,
+      "loss": 0.1074,
+      "step": 1578
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012622997964456087,
+      "loss": 0.1671,
+      "step": 1579
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012614968470980403,
+      "loss": 0.167,
+      "step": 1580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012606937167406227,
+      "loss": 0.301,
+      "step": 1581
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012598904059292883,
+      "loss": 0.1217,
+      "step": 1582
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012590869152200943,
+      "loss": 0.1397,
+      "step": 1583
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001258283245169222,
+      "loss": 0.137,
+      "step": 1584
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012574793963329777,
+      "loss": 0.144,
+      "step": 1585
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000125667536926779,
+      "loss": 0.1195,
+      "step": 1586
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012558711645302125,
+      "loss": 0.2136,
+      "step": 1587
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012550667826769207,
+      "loss": 0.1265,
+      "step": 1588
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012542622242647133,
+      "loss": 0.1397,
+      "step": 1589
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012534574898505105,
+      "loss": 0.204,
+      "step": 1590
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001252652579991355,
+      "loss": 0.1662,
+      "step": 1591
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001251847495244411,
+      "loss": 0.1531,
+      "step": 1592
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012510422361669633,
+      "loss": 0.1859,
+      "step": 1593
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012502368033164176,
+      "loss": 0.1279,
+      "step": 1594
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012494311972502994,
+      "loss": 0.1303,
+      "step": 1595
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012486254185262554,
+      "loss": 0.103,
+      "step": 1596
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012478194677020504,
+      "loss": 0.1024,
+      "step": 1597
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001247013345335569,
+      "loss": 0.2223,
+      "step": 1598
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012462070519848146,
+      "loss": 0.1945,
+      "step": 1599
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012454005882079082,
+      "loss": 0.1192,
+      "step": 1600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012445939545630902,
+      "loss": 0.113,
+      "step": 1601
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012437871516087173,
+      "loss": 0.3204,
+      "step": 1602
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001242980179903264,
+      "loss": 0.205,
+      "step": 1603
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012421730400053214,
+      "loss": 0.1563,
+      "step": 1604
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001241365732473597,
+      "loss": 0.1735,
+      "step": 1605
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012405582578669145,
+      "loss": 0.1196,
+      "step": 1606
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001239750616744213,
+      "loss": 0.1918,
+      "step": 1607
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012389428096645474,
+      "loss": 0.1386,
+      "step": 1608
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012381348371870866,
+      "loss": 0.1293,
+      "step": 1609
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001237326699871115,
+      "loss": 0.0878,
+      "step": 1610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012365183982760304,
+      "loss": 0.126,
+      "step": 1611
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012357099329613443,
+      "loss": 0.153,
+      "step": 1612
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012349013044866822,
+      "loss": 0.2073,
+      "step": 1613
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012340925134117817,
+      "loss": 0.1982,
+      "step": 1614
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012332835602964933,
+      "loss": 0.067,
+      "step": 1615
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012324744457007804,
+      "loss": 0.1066,
+      "step": 1616
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012316651701847165,
+      "loss": 0.1575,
+      "step": 1617
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001230855734308488,
+      "loss": 0.0667,
+      "step": 1618
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012300461386323915,
+      "loss": 0.1103,
+      "step": 1619
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012292363837168352,
+      "loss": 0.1956,
+      "step": 1620
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001228426470122336,
+      "loss": 0.1939,
+      "step": 1621
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012276163984095218,
+      "loss": 0.1429,
+      "step": 1622
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012268061691391298,
+      "loss": 0.2205,
+      "step": 1623
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012259957828720052,
+      "loss": 0.0752,
+      "step": 1624
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012251852401691038,
+      "loss": 0.0957,
+      "step": 1625
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012243745415914883,
+      "loss": 0.238,
+      "step": 1626
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001223563687700329,
+      "loss": 0.1446,
+      "step": 1627
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012227526790569053,
+      "loss": 0.1751,
+      "step": 1628
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001221941516222602,
+      "loss": 0.1504,
+      "step": 1629
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001221130199758912,
+      "loss": 0.0831,
+      "step": 1630
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012203187302274334,
+      "loss": 0.1503,
+      "step": 1631
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001219507108189871,
+      "loss": 0.0945,
+      "step": 1632
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001218695334208035,
+      "loss": 0.1743,
+      "step": 1633
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012178834088438404,
+      "loss": 0.0649,
+      "step": 1634
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001217071332659308,
+      "loss": 0.1213,
+      "step": 1635
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012162591062165615,
+      "loss": 0.0768,
+      "step": 1636
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00012154467300778301,
+      "loss": 0.1114,
+      "step": 1637
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012146342048054453,
+      "loss": 0.1131,
+      "step": 1638
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012138215309618433,
+      "loss": 0.0934,
+      "step": 1639
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012130087091095617,
+      "loss": 0.1513,
+      "step": 1640
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012121957398112411,
+      "loss": 0.1658,
+      "step": 1641
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012113826236296244,
+      "loss": 0.0833,
+      "step": 1642
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012105693611275562,
+      "loss": 0.1723,
+      "step": 1643
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012097559528679821,
+      "loss": 0.1107,
+      "step": 1644
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012089423994139484,
+      "loss": 0.0762,
+      "step": 1645
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012081287013286025,
+      "loss": 0.1374,
+      "step": 1646
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012073148591751915,
+      "loss": 0.1672,
+      "step": 1647
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012065008735170624,
+      "loss": 0.1964,
+      "step": 1648
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012056867449176614,
+      "loss": 0.2148,
+      "step": 1649
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012048724739405338,
+      "loss": 0.2328,
+      "step": 1650
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012040580611493229,
+      "loss": 0.0594,
+      "step": 1651
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012032435071077712,
+      "loss": 0.1178,
+      "step": 1652
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001202428812379718,
+      "loss": 0.1606,
+      "step": 1653
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001201613977529101,
+      "loss": 0.2155,
+      "step": 1654
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012007990031199534,
+      "loss": 0.0694,
+      "step": 1655
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011999838897164066,
+      "loss": 0.1084,
+      "step": 1656
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011991686378826869,
+      "loss": 0.1289,
+      "step": 1657
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001198353248183118,
+      "loss": 0.138,
+      "step": 1658
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011975377211821168,
+      "loss": 0.1323,
+      "step": 1659
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011967220574441973,
+      "loss": 0.1091,
+      "step": 1660
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011959062575339673,
+      "loss": 0.2315,
+      "step": 1661
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011950903220161285,
+      "loss": 0.2365,
+      "step": 1662
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001194274251455477,
+      "loss": 0.2572,
+      "step": 1663
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011934580464169025,
+      "loss": 0.1066,
+      "step": 1664
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011926417074653872,
+      "loss": 0.2786,
+      "step": 1665
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011918252351660066,
+      "loss": 0.2567,
+      "step": 1666
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011910086300839282,
+      "loss": 0.1806,
+      "step": 1667
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011901918927844114,
+      "loss": 0.137,
+      "step": 1668
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011893750238328072,
+      "loss": 0.1901,
+      "step": 1669
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011885580237945572,
+      "loss": 0.06,
+      "step": 1670
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011877408932351953,
+      "loss": 0.153,
+      "step": 1671
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011869236327203434,
+      "loss": 0.205,
+      "step": 1672
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011861062428157156,
+      "loss": 0.159,
+      "step": 1673
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011852887240871145,
+      "loss": 0.1489,
+      "step": 1674
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011844710771004315,
+      "loss": 0.1525,
+      "step": 1675
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011836533024216476,
+      "loss": 0.1809,
+      "step": 1676
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011828354006168318,
+      "loss": 0.0825,
+      "step": 1677
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011820173722521413,
+      "loss": 0.2067,
+      "step": 1678
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011811992178938203,
+      "loss": 0.1473,
+      "step": 1679
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001180380938108201,
+      "loss": 0.1131,
+      "step": 1680
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001179562533461702,
+      "loss": 0.0838,
+      "step": 1681
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011787440045208286,
+      "loss": 0.1138,
+      "step": 1682
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011779253518521716,
+      "loss": 0.1128,
+      "step": 1683
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001177106576022408,
+      "loss": 0.1019,
+      "step": 1684
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011762876775983,
+      "loss": 0.1055,
+      "step": 1685
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011754686571466944,
+      "loss": 0.2101,
+      "step": 1686
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011746495152345223,
+      "loss": 0.2123,
+      "step": 1687
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011738302524287997,
+      "loss": 0.137,
+      "step": 1688
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001173010869296625,
+      "loss": 0.1267,
+      "step": 1689
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011721913664051813,
+      "loss": 0.1773,
+      "step": 1690
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011713717443217336,
+      "loss": 0.1089,
+      "step": 1691
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000117055200361363,
+      "loss": 0.1546,
+      "step": 1692
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011697321448482998,
+      "loss": 0.189,
+      "step": 1693
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011689121685932556,
+      "loss": 0.1515,
+      "step": 1694
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.000116809207541609,
+      "loss": 0.0684,
+      "step": 1695
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011672718658844765,
+      "loss": 0.1203,
+      "step": 1696
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011664515405661698,
+      "loss": 0.0797,
+      "step": 1697
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001165631100029005,
+      "loss": 0.2115,
+      "step": 1698
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011648105448408962,
+      "loss": 0.1844,
+      "step": 1699
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011639898755698366,
+      "loss": 0.0655,
+      "step": 1700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011631690927838994,
+      "loss": 0.2228,
+      "step": 1701
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001162348197051236,
+      "loss": 0.1263,
+      "step": 1702
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011615271889400758,
+      "loss": 0.1677,
+      "step": 1703
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011607060690187257,
+      "loss": 0.1644,
+      "step": 1704
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011598848378555707,
+      "loss": 0.1485,
+      "step": 1705
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011590634960190721,
+      "loss": 0.1843,
+      "step": 1706
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011582420440777686,
+      "loss": 0.1434,
+      "step": 1707
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011574204826002743,
+      "loss": 0.0558,
+      "step": 1708
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.000115659881215528,
+      "loss": 0.1713,
+      "step": 1709
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011557770333115506,
+      "loss": 0.1873,
+      "step": 1710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011549551466379276,
+      "loss": 0.1318,
+      "step": 1711
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011541331527033263,
+      "loss": 0.1211,
+      "step": 1712
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011533110520767358,
+      "loss": 0.2328,
+      "step": 1713
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000115248884532722,
+      "loss": 0.1375,
+      "step": 1714
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001151666533023916,
+      "loss": 0.1278,
+      "step": 1715
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011508441157360336,
+      "loss": 0.2127,
+      "step": 1716
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011500215940328556,
+      "loss": 0.2144,
+      "step": 1717
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011491989684837366,
+      "loss": 0.1608,
+      "step": 1718
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011483762396581044,
+      "loss": 0.293,
+      "step": 1719
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011475534081254562,
+      "loss": 0.1608,
+      "step": 1720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011467304744553618,
+      "loss": 0.1171,
+      "step": 1721
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011459074392174618,
+      "loss": 0.1557,
+      "step": 1722
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001145084302981466,
+      "loss": 0.1111,
+      "step": 1723
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011442610663171553,
+      "loss": 0.1082,
+      "step": 1724
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011434377297943793,
+      "loss": 0.1116,
+      "step": 1725
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011426142939830571,
+      "loss": 0.1052,
+      "step": 1726
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001141790759453176,
+      "loss": 0.113,
+      "step": 1727
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011409671267747927,
+      "loss": 0.1554,
+      "step": 1728
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001140143396518031,
+      "loss": 0.0897,
+      "step": 1729
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011393195692530823,
+      "loss": 0.2029,
+      "step": 1730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011384956455502049,
+      "loss": 0.1545,
+      "step": 1731
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001137671625979725,
+      "loss": 0.179,
+      "step": 1732
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011368475111120343,
+      "loss": 0.0812,
+      "step": 1733
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011360233015175899,
+      "loss": 0.166,
+      "step": 1734
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011351989977669156,
+      "loss": 0.1095,
+      "step": 1735
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011343746004306002,
+      "loss": 0.2102,
+      "step": 1736
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011335501100792962,
+      "loss": 0.0615,
+      "step": 1737
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011327255272837221,
+      "loss": 0.1674,
+      "step": 1738
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011319008526146593,
+      "loss": 0.1273,
+      "step": 1739
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011310760866429529,
+      "loss": 0.12,
+      "step": 1740
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011302512299395112,
+      "loss": 0.1327,
+      "step": 1741
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011294262830753057,
+      "loss": 0.1251,
+      "step": 1742
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011286012466213705,
+      "loss": 0.1302,
+      "step": 1743
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011277761211488007,
+      "loss": 0.1249,
+      "step": 1744
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011269509072287538,
+      "loss": 0.1082,
+      "step": 1745
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011261256054324488,
+      "loss": 0.1964,
+      "step": 1746
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011253002163311641,
+      "loss": 0.157,
+      "step": 1747
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011244747404962401,
+      "loss": 0.2006,
+      "step": 1748
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011236491784990767,
+      "loss": 0.1527,
+      "step": 1749
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011228235309111332,
+      "loss": 0.1412,
+      "step": 1750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011219977983039285,
+      "loss": 0.0657,
+      "step": 1751
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011211719812490397,
+      "loss": 0.0959,
+      "step": 1752
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011203460803181038,
+      "loss": 0.0869,
+      "step": 1753
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011195200960828139,
+      "loss": 0.0696,
+      "step": 1754
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001118694029114922,
+      "loss": 0.1908,
+      "step": 1755
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011178678799862377,
+      "loss": 0.2232,
+      "step": 1756
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011170416492686263,
+      "loss": 0.216,
+      "step": 1757
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011162153375340101,
+      "loss": 0.1204,
+      "step": 1758
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011153889453543676,
+      "loss": 0.1785,
+      "step": 1759
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011145624733017338,
+      "loss": 0.2871,
+      "step": 1760
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011137359219481969,
+      "loss": 0.2227,
+      "step": 1761
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011129092918659019,
+      "loss": 0.1612,
+      "step": 1762
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011120825836270476,
+      "loss": 0.1214,
+      "step": 1763
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011112557978038866,
+      "loss": 0.0823,
+      "step": 1764
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001110428934968726,
+      "loss": 0.2006,
+      "step": 1765
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011096019956939254,
+      "loss": 0.0933,
+      "step": 1766
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011087749805518977,
+      "loss": 0.1333,
+      "step": 1767
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011079478901151081,
+      "loss": 0.1845,
+      "step": 1768
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011071207249560742,
+      "loss": 0.1276,
+      "step": 1769
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011062934856473655,
+      "loss": 0.0901,
+      "step": 1770
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011054661727616022,
+      "loss": 0.08,
+      "step": 1771
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011046387868714556,
+      "loss": 0.2028,
+      "step": 1772
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011038113285496481,
+      "loss": 0.1855,
+      "step": 1773
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011029837983689517,
+      "loss": 0.1542,
+      "step": 1774
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011021561969021879,
+      "loss": 0.1194,
+      "step": 1775
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011013285247222283,
+      "loss": 0.0821,
+      "step": 1776
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011005007824019933,
+      "loss": 0.1163,
+      "step": 1777
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001099672970514451,
+      "loss": 0.1966,
+      "step": 1778
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00010988450896326186,
+      "loss": 0.1742,
+      "step": 1779
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001098017140329561,
+      "loss": 0.121,
+      "step": 1780
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00010971891231783893,
+      "loss": 0.1411,
+      "step": 1781
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00010963610387522631,
+      "loss": 0.1025,
+      "step": 1782
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00010955328876243885,
+      "loss": 0.1493,
+      "step": 1783
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00010947046703680161,
+      "loss": 0.1968,
+      "step": 1784
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00010938763875564437,
+      "loss": 0.1487,
+      "step": 1785
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00010930480397630145,
+      "loss": 0.0742,
+      "step": 1786
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00010922196275611161,
+      "loss": 0.1701,
+      "step": 1787
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001091391151524181,
+      "loss": 0.1315,
+      "step": 1788
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00010905626122256855,
+      "loss": 0.072,
+      "step": 1789
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00010897340102391505,
+      "loss": 0.1369,
+      "step": 1790
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00010889053461381391,
+      "loss": 0.1499,
+      "step": 1791
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00010880766204962585,
+      "loss": 0.3475,
+      "step": 1792
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001087247833887158,
+      "loss": 0.0397,
+      "step": 1793
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00010864189868845297,
+      "loss": 0.1144,
+      "step": 1794
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00010855900800621062,
+      "loss": 0.1164,
+      "step": 1795
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00010847611139936622,
+      "loss": 0.1561,
+      "step": 1796
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001083932089253014,
+      "loss": 0.182,
+      "step": 1797
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00010831030064140184,
+      "loss": 0.1303,
+      "step": 1798
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00010822738660505714,
+      "loss": 0.2402,
+      "step": 1799
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001081444668736609,
+      "loss": 0.2039,
+      "step": 1800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00010806154150461083,
+      "loss": 0.1847,
+      "step": 1801
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00010797861055530831,
+      "loss": 0.1169,
+      "step": 1802
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00010789567408315875,
+      "loss": 0.1602,
+      "step": 1803
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00010781273214557133,
+      "loss": 0.149,
+      "step": 1804
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00010772978479995899,
+      "loss": 0.179,
+      "step": 1805
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001076468321037384,
+      "loss": 0.1093,
+      "step": 1806
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00010756387411433,
+      "loss": 0.143,
+      "step": 1807
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001074809108891579,
+      "loss": 0.2171,
+      "step": 1808
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010739794248564972,
+      "loss": 0.15,
+      "step": 1809
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010731496896123676,
+      "loss": 0.0763,
+      "step": 1810
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001072319903733539,
+      "loss": 0.1037,
+      "step": 1811
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010714900677943938,
+      "loss": 0.0848,
+      "step": 1812
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010706601823693503,
+      "loss": 0.126,
+      "step": 1813
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010698302480328605,
+      "loss": 0.0759,
+      "step": 1814
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010690002653594114,
+      "loss": 0.1586,
+      "step": 1815
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010681702349235212,
+      "loss": 0.1472,
+      "step": 1816
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001067340157299743,
+      "loss": 0.1111,
+      "step": 1817
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010665100330626625,
+      "loss": 0.1259,
+      "step": 1818
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010656798627868963,
+      "loss": 0.2843,
+      "step": 1819
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001064849647047094,
+      "loss": 0.1387,
+      "step": 1820
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010640193864179373,
+      "loss": 0.1067,
+      "step": 1821
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010631890814741372,
+      "loss": 0.1479,
+      "step": 1822
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010623587327904363,
+      "loss": 0.1789,
+      "step": 1823
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010615283409416077,
+      "loss": 0.0879,
+      "step": 1824
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010606979065024547,
+      "loss": 0.1307,
+      "step": 1825
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010598674300478084,
+      "loss": 0.0865,
+      "step": 1826
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010590369121525305,
+      "loss": 0.1207,
+      "step": 1827
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010582063533915114,
+      "loss": 0.1487,
+      "step": 1828
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010573757543396692,
+      "loss": 0.0681,
+      "step": 1829
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010565451155719496,
+      "loss": 0.1921,
+      "step": 1830
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010557144376633268,
+      "loss": 0.2491,
+      "step": 1831
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010548837211888015,
+      "loss": 0.1251,
+      "step": 1832
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001054052966723401,
+      "loss": 0.2788,
+      "step": 1833
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010532221748421787,
+      "loss": 0.0796,
+      "step": 1834
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010523913461202148,
+      "loss": 0.1187,
+      "step": 1835
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010515604811326143,
+      "loss": 0.1425,
+      "step": 1836
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010507295804545074,
+      "loss": 0.1588,
+      "step": 1837
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010498986446610488,
+      "loss": 0.1257,
+      "step": 1838
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010490676743274181,
+      "loss": 0.1565,
+      "step": 1839
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010482366700288184,
+      "loss": 0.09,
+      "step": 1840
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010474056323404762,
+      "loss": 0.202,
+      "step": 1841
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010465745618376417,
+      "loss": 0.0654,
+      "step": 1842
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010457434590955868,
+      "loss": 0.128,
+      "step": 1843
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010449123246896067,
+      "loss": 0.0515,
+      "step": 1844
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010440811591950182,
+      "loss": 0.0764,
+      "step": 1845
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010432499631871595,
+      "loss": 0.0944,
+      "step": 1846
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010424187372413896,
+      "loss": 0.0883,
+      "step": 1847
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001041587481933089,
+      "loss": 0.1375,
+      "step": 1848
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010407561978376584,
+      "loss": 0.1641,
+      "step": 1849
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010399248855305176,
+      "loss": 0.099,
+      "step": 1850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010390935455871068,
+      "loss": 0.1836,
+      "step": 1851
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010382621785828848,
+      "loss": 0.2102,
+      "step": 1852
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010374307850933293,
+      "loss": 0.1791,
+      "step": 1853
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010365993656939367,
+      "loss": 0.1498,
+      "step": 1854
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010357679209602207,
+      "loss": 0.0655,
+      "step": 1855
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010349364514677131,
+      "loss": 0.1723,
+      "step": 1856
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010341049577919624,
+      "loss": 0.2157,
+      "step": 1857
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010332734405085342,
+      "loss": 0.0909,
+      "step": 1858
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010324419001930104,
+      "loss": 0.106,
+      "step": 1859
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001031610337420988,
+      "loss": 0.0797,
+      "step": 1860
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001030778752768081,
+      "loss": 0.1445,
+      "step": 1861
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010299471468099175,
+      "loss": 0.0834,
+      "step": 1862
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010291155201221407,
+      "loss": 0.0835,
+      "step": 1863
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010282838732804081,
+      "loss": 0.0766,
+      "step": 1864
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010274522068603911,
+      "loss": 0.1143,
+      "step": 1865
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010266205214377748,
+      "loss": 0.1251,
+      "step": 1866
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010257888175882575,
+      "loss": 0.0818,
+      "step": 1867
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010249570958875496,
+      "loss": 0.1258,
+      "step": 1868
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001024125356911375,
+      "loss": 0.2629,
+      "step": 1869
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010232936012354682,
+      "loss": 0.1288,
+      "step": 1870
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010224618294355768,
+      "loss": 0.1566,
+      "step": 1871
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010216300420874578,
+      "loss": 0.1149,
+      "step": 1872
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010207982397668809,
+      "loss": 0.1949,
+      "step": 1873
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010199664230496246,
+      "loss": 0.1378,
+      "step": 1874
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010191345925114781,
+      "loss": 0.1126,
+      "step": 1875
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010183027487282401,
+      "loss": 0.1277,
+      "step": 1876
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010174708922757181,
+      "loss": 0.1663,
+      "step": 1877
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010166390237297287,
+      "loss": 0.0978,
+      "step": 1878
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010158071436660972,
+      "loss": 0.1918,
+      "step": 1879
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001014975252660656,
+      "loss": 0.1256,
+      "step": 1880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010141433512892457,
+      "loss": 0.1501,
+      "step": 1881
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010133114401277139,
+      "loss": 0.1142,
+      "step": 1882
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010124795197519152,
+      "loss": 0.2173,
+      "step": 1883
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010116475907377105,
+      "loss": 0.0916,
+      "step": 1884
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010108156536609663,
+      "loss": 0.1124,
+      "step": 1885
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001009983709097555,
+      "loss": 0.1679,
+      "step": 1886
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001009151757623354,
+      "loss": 0.1314,
+      "step": 1887
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010083197998142459,
+      "loss": 0.0787,
+      "step": 1888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010074878362461172,
+      "loss": 0.0675,
+      "step": 1889
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001006655867494859,
+      "loss": 0.0938,
+      "step": 1890
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010058238941363646,
+      "loss": 0.1692,
+      "step": 1891
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010049919167465327,
+      "loss": 0.2454,
+      "step": 1892
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010041599359012628,
+      "loss": 0.1315,
+      "step": 1893
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 3786,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 1893,
+  "total_flos": 1.0710634393371673e+19,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}