diff --git "a/checkpoint-1000/trainer_state.json" "b/checkpoint-1000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1000/trainer_state.json"
@@ -0,0 +1,6019 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.8368794326241136,
+  "eval_steps": 500,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019981060606060605,
+      "loss": 2.9206,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019962121212121212,
+      "loss": 2.7609,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001994318181818182,
+      "loss": 2.6878,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019924242424242426,
+      "loss": 2.6697,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001990530303030303,
+      "loss": 2.5818,
+      "step": 5
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019886363636363637,
+      "loss": 2.5396,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019867424242424244,
+      "loss": 2.5265,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001984848484848485,
+      "loss": 2.5475,
+      "step": 8
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019829545454545455,
+      "loss": 2.4835,
+      "step": 9
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001981060606060606,
+      "loss": 2.4559,
+      "step": 10
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001979166666666667,
+      "loss": 2.4511,
+      "step": 11
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019772727272727273,
+      "loss": 2.4592,
+      "step": 12
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001975378787878788,
+      "loss": 2.4495,
+      "step": 13
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019734848484848484,
+      "loss": 2.4714,
+      "step": 14
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019715909090909094,
+      "loss": 2.4302,
+      "step": 15
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019696969696969698,
+      "loss": 2.4097,
+      "step": 16
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019678030303030305,
+      "loss": 2.4523,
+      "step": 17
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001965909090909091,
+      "loss": 2.4325,
+      "step": 18
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019640151515151516,
+      "loss": 2.4125,
+      "step": 19
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019621212121212123,
+      "loss": 2.4329,
+      "step": 20
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019602272727272727,
+      "loss": 2.3471,
+      "step": 21
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019583333333333334,
+      "loss": 2.3012,
+      "step": 22
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001956439393939394,
+      "loss": 2.3869,
+      "step": 23
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019545454545454548,
+      "loss": 2.3822,
+      "step": 24
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019526515151515152,
+      "loss": 2.3427,
+      "step": 25
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001950757575757576,
+      "loss": 2.3659,
+      "step": 26
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019488636363636366,
+      "loss": 2.3826,
+      "step": 27
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001946969696969697,
+      "loss": 2.3532,
+      "step": 28
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019450757575757577,
+      "loss": 2.3828,
+      "step": 29
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001943181818181818,
+      "loss": 2.3133,
+      "step": 30
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001941287878787879,
+      "loss": 2.3613,
+      "step": 31
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019393939393939395,
+      "loss": 2.3867,
+      "step": 32
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019375000000000002,
+      "loss": 2.2966,
+      "step": 33
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019356060606060606,
+      "loss": 2.3436,
+      "step": 34
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019337121212121213,
+      "loss": 2.3425,
+      "step": 35
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001931818181818182,
+      "loss": 2.307,
+      "step": 36
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019299242424242424,
+      "loss": 2.3521,
+      "step": 37
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001928030303030303,
+      "loss": 2.3302,
+      "step": 38
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019261363636363635,
+      "loss": 2.312,
+      "step": 39
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019242424242424245,
+      "loss": 2.3655,
+      "step": 40
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001922348484848485,
+      "loss": 2.344,
+      "step": 41
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019204545454545456,
+      "loss": 2.3373,
+      "step": 42
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001918560606060606,
+      "loss": 2.3331,
+      "step": 43
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019166666666666667,
+      "loss": 2.3376,
+      "step": 44
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019147727272727274,
+      "loss": 2.3369,
+      "step": 45
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019128787878787878,
+      "loss": 2.3413,
+      "step": 46
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019109848484848485,
+      "loss": 2.3212,
+      "step": 47
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019090909090909092,
+      "loss": 2.307,
+      "step": 48
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000190719696969697,
+      "loss": 2.2929,
+      "step": 49
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019053030303030303,
+      "loss": 2.2873,
+      "step": 50
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001903409090909091,
+      "loss": 2.3098,
+      "step": 51
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019015151515151517,
+      "loss": 2.3129,
+      "step": 52
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001899621212121212,
+      "loss": 2.3038,
+      "step": 53
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018977272727272728,
+      "loss": 2.286,
+      "step": 54
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018958333333333332,
+      "loss": 2.3388,
+      "step": 55
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018939393939393942,
+      "loss": 2.3193,
+      "step": 56
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018920454545454546,
+      "loss": 2.3136,
+      "step": 57
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018901515151515153,
+      "loss": 2.3141,
+      "step": 58
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018882575757575757,
+      "loss": 2.3646,
+      "step": 59
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018863636363636364,
+      "loss": 2.3318,
+      "step": 60
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001884469696969697,
+      "loss": 2.2977,
+      "step": 61
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018825757575757575,
+      "loss": 2.2764,
+      "step": 62
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018806818181818182,
+      "loss": 2.3095,
+      "step": 63
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001878787878787879,
+      "loss": 2.252,
+      "step": 64
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018768939393939396,
+      "loss": 2.2786,
+      "step": 65
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001875,
+      "loss": 2.2789,
+      "step": 66
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018731060606060607,
+      "loss": 2.2841,
+      "step": 67
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018712121212121212,
+      "loss": 2.3436,
+      "step": 68
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018693181818181818,
+      "loss": 2.2956,
+      "step": 69
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018674242424242425,
+      "loss": 2.2353,
+      "step": 70
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001865530303030303,
+      "loss": 2.2772,
+      "step": 71
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018636363636363636,
+      "loss": 2.2496,
+      "step": 72
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018617424242424243,
+      "loss": 2.2477,
+      "step": 73
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001859848484848485,
+      "loss": 2.2791,
+      "step": 74
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018579545454545454,
+      "loss": 2.2799,
+      "step": 75
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018560606060606061,
+      "loss": 2.3132,
+      "step": 76
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018541666666666668,
+      "loss": 2.2542,
+      "step": 77
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018522727272727273,
+      "loss": 2.2609,
+      "step": 78
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001850378787878788,
+      "loss": 2.2819,
+      "step": 79
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018484848484848484,
+      "loss": 2.2844,
+      "step": 80
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018465909090909093,
+      "loss": 2.2542,
+      "step": 81
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018446969696969697,
+      "loss": 2.2603,
+      "step": 82
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018428030303030304,
+      "loss": 2.2832,
+      "step": 83
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018409090909090909,
+      "loss": 2.2869,
+      "step": 84
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018390151515151518,
+      "loss": 2.2646,
+      "step": 85
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018371212121212122,
+      "loss": 2.2698,
+      "step": 86
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018352272727272727,
+      "loss": 2.2757,
+      "step": 87
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018333333333333334,
+      "loss": 2.2544,
+      "step": 88
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001831439393939394,
+      "loss": 2.2678,
+      "step": 89
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018295454545454547,
+      "loss": 2.2778,
+      "step": 90
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018276515151515152,
+      "loss": 2.2027,
+      "step": 91
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018257575757575758,
+      "loss": 2.2167,
+      "step": 92
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018238636363636365,
+      "loss": 2.2602,
+      "step": 93
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018219696969696972,
+      "loss": 2.2736,
+      "step": 94
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018200757575757577,
+      "loss": 2.2443,
+      "step": 95
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018181818181818183,
+      "loss": 2.2299,
+      "step": 96
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001816287878787879,
+      "loss": 2.2644,
+      "step": 97
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018143939393939395,
+      "loss": 2.259,
+      "step": 98
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018125000000000001,
+      "loss": 2.2567,
+      "step": 99
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018106060606060606,
+      "loss": 2.2599,
+      "step": 100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018087121212121213,
+      "loss": 2.2091,
+      "step": 101
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001806818181818182,
+      "loss": 2.2312,
+      "step": 102
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018049242424242426,
+      "loss": 2.1869,
+      "step": 103
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001803030303030303,
+      "loss": 2.2023,
+      "step": 104
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018011363636363638,
+      "loss": 2.2132,
+      "step": 105
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017992424242424244,
+      "loss": 2.2612,
+      "step": 106
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001797348484848485,
+      "loss": 2.2109,
+      "step": 107
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017954545454545456,
+      "loss": 2.215,
+      "step": 108
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001793560606060606,
+      "loss": 2.2114,
+      "step": 109
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001791666666666667,
+      "loss": 2.2203,
+      "step": 110
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017897727272727274,
+      "loss": 2.2594,
+      "step": 111
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001787878787878788,
+      "loss": 2.2001,
+      "step": 112
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017859848484848485,
+      "loss": 2.2046,
+      "step": 113
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017840909090909092,
+      "loss": 2.1907,
+      "step": 114
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017821969696969699,
+      "loss": 2.2539,
+      "step": 115
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017803030303030303,
+      "loss": 2.2335,
+      "step": 116
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001778409090909091,
+      "loss": 2.2171,
+      "step": 117
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017765151515151517,
+      "loss": 2.2278,
+      "step": 118
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017746212121212123,
+      "loss": 2.231,
+      "step": 119
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017727272727272728,
+      "loss": 2.2141,
+      "step": 120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017708333333333335,
+      "loss": 2.2432,
+      "step": 121
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017689393939393942,
+      "loss": 2.2266,
+      "step": 122
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017670454545454546,
+      "loss": 2.1929,
+      "step": 123
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017651515151515153,
+      "loss": 2.2077,
+      "step": 124
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017632575757575757,
+      "loss": 2.2133,
+      "step": 125
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017613636363636366,
+      "loss": 2.2251,
+      "step": 126
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001759469696969697,
+      "loss": 2.2265,
+      "step": 127
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017575757575757578,
+      "loss": 2.2186,
+      "step": 128
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017556818181818182,
+      "loss": 2.1925,
+      "step": 129
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001753787878787879,
+      "loss": 2.1956,
+      "step": 130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017518939393939396,
+      "loss": 2.2459,
+      "step": 131
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000175,
+      "loss": 2.22,
+      "step": 132
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017481060606060607,
+      "loss": 2.2143,
+      "step": 133
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001746212121212121,
+      "loss": 2.2359,
+      "step": 134
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001744318181818182,
+      "loss": 2.2058,
+      "step": 135
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017424242424242425,
+      "loss": 2.2307,
+      "step": 136
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017405303030303032,
+      "loss": 2.2062,
+      "step": 137
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017386363636363636,
+      "loss": 2.1796,
+      "step": 138
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017367424242424243,
+      "loss": 2.2054,
+      "step": 139
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001734848484848485,
+      "loss": 2.1651,
+      "step": 140
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017329545454545454,
+      "loss": 2.2159,
+      "step": 141
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001731060606060606,
+      "loss": 2.1988,
+      "step": 142
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017291666666666668,
+      "loss": 2.1676,
+      "step": 143
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017272727272727275,
+      "loss": 2.1725,
+      "step": 144
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001725378787878788,
+      "loss": 2.2205,
+      "step": 145
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017234848484848486,
+      "loss": 2.1486,
+      "step": 146
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017215909090909093,
+      "loss": 2.147,
+      "step": 147
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017196969696969697,
+      "loss": 2.1651,
+      "step": 148
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017178030303030304,
+      "loss": 2.1983,
+      "step": 149
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017159090909090908,
+      "loss": 2.1778,
+      "step": 150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017140151515151518,
+      "loss": 2.1631,
+      "step": 151
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017121212121212122,
+      "loss": 2.1442,
+      "step": 152
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001710227272727273,
+      "loss": 2.1397,
+      "step": 153
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017083333333333333,
+      "loss": 2.1697,
+      "step": 154
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001706439393939394,
+      "loss": 2.1451,
+      "step": 155
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017045454545454547,
+      "loss": 2.1789,
+      "step": 156
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001702651515151515,
+      "loss": 2.1037,
+      "step": 157
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017007575757575758,
+      "loss": 2.1698,
+      "step": 158
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016988636363636365,
+      "loss": 2.1538,
+      "step": 159
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016969696969696972,
+      "loss": 2.2015,
+      "step": 160
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016950757575757576,
+      "loss": 2.179,
+      "step": 161
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016931818181818183,
+      "loss": 2.1766,
+      "step": 162
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001691287878787879,
+      "loss": 2.1646,
+      "step": 163
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016893939393939394,
+      "loss": 2.1694,
+      "step": 164
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016875,
+      "loss": 2.1562,
+      "step": 165
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016856060606060605,
+      "loss": 2.1551,
+      "step": 166
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016837121212121212,
+      "loss": 2.1652,
+      "step": 167
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001681818181818182,
+      "loss": 2.1594,
+      "step": 168
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016799242424242426,
+      "loss": 2.1674,
+      "step": 169
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001678030303030303,
+      "loss": 2.1378,
+      "step": 170
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016761363636363637,
+      "loss": 2.1447,
+      "step": 171
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016742424242424244,
+      "loss": 2.1451,
+      "step": 172
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016723484848484848,
+      "loss": 2.1336,
+      "step": 173
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016704545454545455,
+      "loss": 2.1231,
+      "step": 174
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001668560606060606,
+      "loss": 2.1143,
+      "step": 175
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001666666666666667,
+      "loss": 2.1316,
+      "step": 176
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016647727272727273,
+      "loss": 2.1281,
+      "step": 177
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001662878787878788,
+      "loss": 2.136,
+      "step": 178
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016609848484848484,
+      "loss": 2.1279,
+      "step": 179
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016590909090909094,
+      "loss": 2.1421,
+      "step": 180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016571969696969698,
+      "loss": 2.1541,
+      "step": 181
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016553030303030305,
+      "loss": 2.1293,
+      "step": 182
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001653409090909091,
+      "loss": 2.1294,
+      "step": 183
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016515151515151516,
+      "loss": 2.1459,
+      "step": 184
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016496212121212123,
+      "loss": 2.1113,
+      "step": 185
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016477272727272727,
+      "loss": 2.1394,
+      "step": 186
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016458333333333334,
+      "loss": 2.1321,
+      "step": 187
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001643939393939394,
+      "loss": 2.148,
+      "step": 188
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016420454545454548,
+      "loss": 2.1631,
+      "step": 189
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016401515151515152,
+      "loss": 2.1276,
+      "step": 190
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001638257575757576,
+      "loss": 2.0706,
+      "step": 191
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016363636363636366,
+      "loss": 2.127,
+      "step": 192
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001634469696969697,
+      "loss": 2.1449,
+      "step": 193
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016325757575757577,
+      "loss": 2.1204,
+      "step": 194
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001630681818181818,
+      "loss": 2.0904,
+      "step": 195
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001628787878787879,
+      "loss": 2.1129,
+      "step": 196
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016268939393939395,
+      "loss": 2.1036,
+      "step": 197
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016250000000000002,
+      "loss": 2.1509,
+      "step": 198
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016231060606060606,
+      "loss": 2.1239,
+      "step": 199
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016212121212121213,
+      "loss": 2.145,
+      "step": 200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001619318181818182,
+      "loss": 2.1221,
+      "step": 201
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016174242424242424,
+      "loss": 2.1181,
+      "step": 202
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001615530303030303,
+      "loss": 2.1306,
+      "step": 203
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016136363636363635,
+      "loss": 2.0199,
+      "step": 204
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016117424242424245,
+      "loss": 2.1178,
+      "step": 205
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001609848484848485,
+      "loss": 2.1584,
+      "step": 206
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016079545454545456,
+      "loss": 2.0872,
+      "step": 207
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001606060606060606,
+      "loss": 2.1033,
+      "step": 208
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016041666666666667,
+      "loss": 2.1381,
+      "step": 209
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016022727272727274,
+      "loss": 2.1127,
+      "step": 210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016003787878787878,
+      "loss": 2.1077,
+      "step": 211
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015984848484848485,
+      "loss": 2.0984,
+      "step": 212
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015965909090909092,
+      "loss": 2.0994,
+      "step": 213
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000159469696969697,
+      "loss": 2.096,
+      "step": 214
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015928030303030303,
+      "loss": 2.0909,
+      "step": 215
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001590909090909091,
+      "loss": 2.118,
+      "step": 216
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015890151515151517,
+      "loss": 2.0783,
+      "step": 217
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001587121212121212,
+      "loss": 2.0876,
+      "step": 218
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015852272727272728,
+      "loss": 2.0581,
+      "step": 219
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015833333333333332,
+      "loss": 2.0548,
+      "step": 220
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015814393939393942,
+      "loss": 2.0595,
+      "step": 221
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015795454545454546,
+      "loss": 2.0719,
+      "step": 222
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015776515151515153,
+      "loss": 2.0903,
+      "step": 223
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015757575757575757,
+      "loss": 2.0941,
+      "step": 224
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015738636363636364,
+      "loss": 2.0926,
+      "step": 225
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001571969696969697,
+      "loss": 2.0816,
+      "step": 226
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015700757575757575,
+      "loss": 2.0894,
+      "step": 227
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015681818181818182,
+      "loss": 2.0798,
+      "step": 228
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001566287878787879,
+      "loss": 2.0672,
+      "step": 229
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015643939393939396,
+      "loss": 2.0787,
+      "step": 230
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015625,
+      "loss": 2.0611,
+      "step": 231
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015606060606060607,
+      "loss": 2.0805,
+      "step": 232
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015587121212121211,
+      "loss": 2.053,
+      "step": 233
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015568181818181818,
+      "loss": 2.0575,
+      "step": 234
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015549242424242425,
+      "loss": 2.0459,
+      "step": 235
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001553030303030303,
+      "loss": 2.0635,
+      "step": 236
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015511363636363636,
+      "loss": 2.0335,
+      "step": 237
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015492424242424243,
+      "loss": 2.0681,
+      "step": 238
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001547348484848485,
+      "loss": 2.0748,
+      "step": 239
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015454545454545454,
+      "loss": 2.1091,
+      "step": 240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001543560606060606,
+      "loss": 2.0732,
+      "step": 241
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015416666666666668,
+      "loss": 2.0746,
+      "step": 242
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015397727272727272,
+      "loss": 2.0306,
+      "step": 243
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001537878787878788,
+      "loss": 2.0864,
+      "step": 244
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015359848484848484,
+      "loss": 2.0664,
+      "step": 245
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015340909090909093,
+      "loss": 2.0801,
+      "step": 246
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015321969696969697,
+      "loss": 2.0799,
+      "step": 247
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015303030303030304,
+      "loss": 2.0621,
+      "step": 248
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015284090909090909,
+      "loss": 2.0687,
+      "step": 249
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015265151515151515,
+      "loss": 2.018,
+      "step": 250
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015246212121212122,
+      "loss": 2.0256,
+      "step": 251
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015227272727272727,
+      "loss": 2.0736,
+      "step": 252
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015208333333333333,
+      "loss": 2.0609,
+      "step": 253
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001518939393939394,
+      "loss": 2.0539,
+      "step": 254
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015170454545454547,
+      "loss": 2.0282,
+      "step": 255
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015151515151515152,
+      "loss": 2.0417,
+      "step": 256
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015132575757575758,
+      "loss": 2.0333,
+      "step": 257
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015113636363636365,
+      "loss": 2.0428,
+      "step": 258
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015094696969696972,
+      "loss": 2.045,
+      "step": 259
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015075757575757576,
+      "loss": 2.0463,
+      "step": 260
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001505681818181818,
+      "loss": 2.0539,
+      "step": 261
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001503787878787879,
+      "loss": 2.0184,
+      "step": 262
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015018939393939394,
+      "loss": 2.0858,
+      "step": 263
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 2.0239,
+      "step": 264
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014981060606060606,
+      "loss": 2.0425,
+      "step": 265
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014962121212121213,
+      "loss": 2.0263,
+      "step": 266
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001494318181818182,
+      "loss": 2.042,
+      "step": 267
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014924242424242426,
+      "loss": 2.026,
+      "step": 268
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001490530303030303,
+      "loss": 2.0411,
+      "step": 269
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014886363636363635,
+      "loss": 2.028,
+      "step": 270
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014867424242424244,
+      "loss": 2.0172,
+      "step": 271
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014848484848484849,
+      "loss": 2.0196,
+      "step": 272
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014829545454545455,
+      "loss": 2.0142,
+      "step": 273
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001481060606060606,
+      "loss": 2.0265,
+      "step": 274
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001479166666666667,
+      "loss": 2.0353,
+      "step": 275
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014772727272727274,
+      "loss": 2.0327,
+      "step": 276
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001475378787878788,
+      "loss": 2.0188,
+      "step": 277
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014734848484848485,
+      "loss": 1.9987,
+      "step": 278
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014715909090909092,
+      "loss": 2.0141,
+      "step": 279
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014696969696969698,
+      "loss": 2.0403,
+      "step": 280
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014678030303030303,
+      "loss": 1.9977,
+      "step": 281
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001465909090909091,
+      "loss": 1.9674,
+      "step": 282
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014640151515151517,
+      "loss": 1.9984,
+      "step": 283
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014621212121212123,
+      "loss": 1.9796,
+      "step": 284
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014602272727272728,
+      "loss": 2.0139,
+      "step": 285
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014583333333333335,
+      "loss": 1.9866,
+      "step": 286
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014564393939393941,
+      "loss": 2.0208,
+      "step": 287
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014545454545454546,
+      "loss": 1.9844,
+      "step": 288
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014526515151515153,
+      "loss": 2.0082,
+      "step": 289
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014507575757575757,
+      "loss": 1.984,
+      "step": 290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014488636363636366,
+      "loss": 2.0015,
+      "step": 291
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001446969696969697,
+      "loss": 2.0209,
+      "step": 292
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014450757575757578,
+      "loss": 1.9728,
+      "step": 293
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014431818181818182,
+      "loss": 2.0032,
+      "step": 294
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014412878787878789,
+      "loss": 1.9641,
+      "step": 295
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014393939393939396,
+      "loss": 1.9945,
+      "step": 296
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014375,
+      "loss": 1.9658,
+      "step": 297
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014356060606060607,
+      "loss": 1.9907,
+      "step": 298
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001433712121212121,
+      "loss": 1.9935,
+      "step": 299
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001431818181818182,
+      "loss": 1.9897,
+      "step": 300
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014299242424242425,
+      "loss": 1.984,
+      "step": 301
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014280303030303032,
+      "loss": 1.9581,
+      "step": 302
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014261363636363636,
+      "loss": 1.9893,
+      "step": 303
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014242424242424243,
+      "loss": 1.9568,
+      "step": 304
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001422348484848485,
+      "loss": 1.98,
+      "step": 305
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014204545454545454,
+      "loss": 1.9519,
+      "step": 306
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001418560606060606,
+      "loss": 1.9693,
+      "step": 307
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014166666666666668,
+      "loss": 1.9866,
+      "step": 308
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014147727272727275,
+      "loss": 1.9508,
+      "step": 309
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001412878787878788,
+      "loss": 1.9653,
+      "step": 310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014109848484848486,
+      "loss": 1.9991,
+      "step": 311
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014090909090909093,
+      "loss": 1.9442,
+      "step": 312
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014071969696969697,
+      "loss": 1.9807,
+      "step": 313
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014053030303030304,
+      "loss": 1.9958,
+      "step": 314
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014034090909090908,
+      "loss": 1.9459,
+      "step": 315
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00014015151515151518,
+      "loss": 1.9508,
+      "step": 316
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00013996212121212122,
+      "loss": 1.9933,
+      "step": 317
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001397727272727273,
+      "loss": 1.9703,
+      "step": 318
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00013958333333333333,
+      "loss": 1.965,
+      "step": 319
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001393939393939394,
+      "loss": 1.9264,
+      "step": 320
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013920454545454547,
+      "loss": 1.9688,
+      "step": 321
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001390151515151515,
+      "loss": 1.9901,
+      "step": 322
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013882575757575758,
+      "loss": 1.9363,
+      "step": 323
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013863636363636365,
+      "loss": 1.9269,
+      "step": 324
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013844696969696972,
+      "loss": 1.9688,
+      "step": 325
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013825757575757576,
+      "loss": 1.9758,
+      "step": 326
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013806818181818183,
+      "loss": 1.9414,
+      "step": 327
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001378787878787879,
+      "loss": 1.9397,
+      "step": 328
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013768939393939394,
+      "loss": 1.9032,
+      "step": 329
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001375,
+      "loss": 1.9777,
+      "step": 330
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013731060606060605,
+      "loss": 1.9173,
+      "step": 331
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013712121212121212,
+      "loss": 1.9307,
+      "step": 332
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001369318181818182,
+      "loss": 1.9611,
+      "step": 333
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013674242424242426,
+      "loss": 1.9698,
+      "step": 334
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001365530303030303,
+      "loss": 1.9619,
+      "step": 335
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013636363636363637,
+      "loss": 1.9322,
+      "step": 336
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013617424242424244,
+      "loss": 1.9441,
+      "step": 337
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013598484848484848,
+      "loss": 1.9563,
+      "step": 338
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013579545454545455,
+      "loss": 1.9283,
+      "step": 339
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001356060606060606,
+      "loss": 1.9508,
+      "step": 340
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001354166666666667,
+      "loss": 1.9285,
+      "step": 341
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013522727272727273,
+      "loss": 1.9295,
+      "step": 342
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001350378787878788,
+      "loss": 1.9272,
+      "step": 343
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013484848484848484,
+      "loss": 1.905,
+      "step": 344
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013465909090909094,
+      "loss": 1.9409,
+      "step": 345
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013446969696969698,
+      "loss": 1.9674,
+      "step": 346
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013428030303030302,
+      "loss": 1.9278,
+      "step": 347
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001340909090909091,
+      "loss": 1.9136,
+      "step": 348
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013390151515151516,
+      "loss": 1.9143,
+      "step": 349
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013371212121212123,
+      "loss": 1.9381,
+      "step": 350
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013352272727272727,
+      "loss": 1.9136,
+      "step": 351
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013333333333333334,
+      "loss": 1.9103,
+      "step": 352
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001331439393939394,
+      "loss": 1.9027,
+      "step": 353
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013295454545454548,
+      "loss": 1.8674,
+      "step": 354
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013276515151515152,
+      "loss": 1.886,
+      "step": 355
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013257575757575756,
+      "loss": 1.887,
+      "step": 356
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013238636363636366,
+      "loss": 1.87,
+      "step": 357
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001321969696969697,
+      "loss": 1.8715,
+      "step": 358
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013200757575757577,
+      "loss": 1.8993,
+      "step": 359
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001318181818181818,
+      "loss": 1.8844,
+      "step": 360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001316287878787879,
+      "loss": 1.8965,
+      "step": 361
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013143939393939395,
+      "loss": 1.8956,
+      "step": 362
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013125000000000002,
+      "loss": 1.869,
+      "step": 363
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013106060606060606,
+      "loss": 1.8702,
+      "step": 364
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013087121212121213,
+      "loss": 1.8962,
+      "step": 365
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001306818181818182,
+      "loss": 1.8613,
+      "step": 366
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013049242424242424,
+      "loss": 1.8845,
+      "step": 367
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001303030303030303,
+      "loss": 1.8689,
+      "step": 368
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00013011363636363635,
+      "loss": 1.9059,
+      "step": 369
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00012992424242424245,
+      "loss": 1.9082,
+      "step": 370
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001297348484848485,
+      "loss": 1.8918,
+      "step": 371
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012954545454545456,
+      "loss": 1.8657,
+      "step": 372
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001293560606060606,
+      "loss": 1.8909,
+      "step": 373
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012916666666666667,
+      "loss": 1.8649,
+      "step": 374
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012897727272727274,
+      "loss": 1.833,
+      "step": 375
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012878787878787878,
+      "loss": 1.8815,
+      "step": 376
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012859848484848485,
+      "loss": 1.8646,
+      "step": 377
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012840909090909092,
+      "loss": 1.846,
+      "step": 378
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.000128219696969697,
+      "loss": 1.8631,
+      "step": 379
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012803030303030303,
+      "loss": 1.917,
+      "step": 380
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001278409090909091,
+      "loss": 1.9068,
+      "step": 381
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012765151515151517,
+      "loss": 1.8772,
+      "step": 382
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001274621212121212,
+      "loss": 1.8414,
+      "step": 383
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012727272727272728,
+      "loss": 1.9003,
+      "step": 384
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012708333333333332,
+      "loss": 1.8415,
+      "step": 385
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012689393939393942,
+      "loss": 1.8491,
+      "step": 386
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012670454545454546,
+      "loss": 1.8875,
+      "step": 387
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012651515151515153,
+      "loss": 1.8629,
+      "step": 388
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012632575757575757,
+      "loss": 1.8378,
+      "step": 389
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012613636363636364,
+      "loss": 1.8442,
+      "step": 390
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001259469696969697,
+      "loss": 1.8587,
+      "step": 391
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012575757575757575,
+      "loss": 1.8659,
+      "step": 392
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012556818181818182,
+      "loss": 1.8271,
+      "step": 393
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001253787878787879,
+      "loss": 1.8692,
+      "step": 394
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012518939393939396,
+      "loss": 1.8071,
+      "step": 395
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000125,
+      "loss": 1.8564,
+      "step": 396
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012481060606060607,
+      "loss": 1.8891,
+      "step": 397
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012462121212121211,
+      "loss": 1.8173,
+      "step": 398
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012443181818181818,
+      "loss": 1.8653,
+      "step": 399
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012424242424242425,
+      "loss": 1.8843,
+      "step": 400
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001240530303030303,
+      "loss": 1.8527,
+      "step": 401
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012386363636363636,
+      "loss": 1.8352,
+      "step": 402
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012367424242424243,
+      "loss": 1.866,
+      "step": 403
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001234848484848485,
+      "loss": 1.8557,
+      "step": 404
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012329545454545454,
+      "loss": 1.8284,
+      "step": 405
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001231060606060606,
+      "loss": 1.8359,
+      "step": 406
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012291666666666668,
+      "loss": 1.8437,
+      "step": 407
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012272727272727272,
+      "loss": 1.8256,
+      "step": 408
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001225378787878788,
+      "loss": 1.8297,
+      "step": 409
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012234848484848484,
+      "loss": 1.8515,
+      "step": 410
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012215909090909093,
+      "loss": 1.8198,
+      "step": 411
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012196969696969697,
+      "loss": 1.7809,
+      "step": 412
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012178030303030303,
+      "loss": 1.8438,
+      "step": 413
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012159090909090908,
+      "loss": 1.8497,
+      "step": 414
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012140151515151517,
+      "loss": 1.8463,
+      "step": 415
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012121212121212122,
+      "loss": 1.7768,
+      "step": 416
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012102272727272728,
+      "loss": 1.8561,
+      "step": 417
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012083333333333333,
+      "loss": 1.863,
+      "step": 418
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001206439393939394,
+      "loss": 1.8193,
+      "step": 419
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012045454545454546,
+      "loss": 1.7732,
+      "step": 420
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012026515151515151,
+      "loss": 1.7728,
+      "step": 421
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012007575757575757,
+      "loss": 1.8113,
+      "step": 422
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011988636363636365,
+      "loss": 1.7976,
+      "step": 423
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011969696969696971,
+      "loss": 1.786,
+      "step": 424
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011950757575757576,
+      "loss": 1.8019,
+      "step": 425
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011931818181818182,
+      "loss": 1.786,
+      "step": 426
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011912878787878789,
+      "loss": 1.8102,
+      "step": 427
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011893939393939394,
+      "loss": 1.7828,
+      "step": 428
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011875,
+      "loss": 1.8498,
+      "step": 429
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011856060606060606,
+      "loss": 1.7983,
+      "step": 430
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011837121212121211,
+      "loss": 1.7863,
+      "step": 431
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001181818181818182,
+      "loss": 1.8171,
+      "step": 432
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011799242424242425,
+      "loss": 1.8143,
+      "step": 433
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001178030303030303,
+      "loss": 1.7815,
+      "step": 434
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011761363636363636,
+      "loss": 1.7652,
+      "step": 435
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011742424242424244,
+      "loss": 1.8242,
+      "step": 436
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011723484848484849,
+      "loss": 1.7789,
+      "step": 437
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011704545454545454,
+      "loss": 1.7549,
+      "step": 438
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001168560606060606,
+      "loss": 1.7528,
+      "step": 439
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011666666666666668,
+      "loss": 1.7443,
+      "step": 440
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011647727272727273,
+      "loss": 1.7911,
+      "step": 441
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011628787878787879,
+      "loss": 1.7848,
+      "step": 442
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011609848484848485,
+      "loss": 1.8137,
+      "step": 443
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011590909090909093,
+      "loss": 1.791,
+      "step": 444
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011571969696969698,
+      "loss": 1.7921,
+      "step": 445
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011553030303030304,
+      "loss": 1.772,
+      "step": 446
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011534090909090908,
+      "loss": 1.776,
+      "step": 447
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011515151515151516,
+      "loss": 1.7948,
+      "step": 448
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011496212121212122,
+      "loss": 1.8187,
+      "step": 449
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011477272727272728,
+      "loss": 1.7436,
+      "step": 450
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011458333333333333,
+      "loss": 1.7326,
+      "step": 451
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011439393939393941,
+      "loss": 1.8005,
+      "step": 452
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011420454545454547,
+      "loss": 1.8088,
+      "step": 453
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011401515151515153,
+      "loss": 1.7632,
+      "step": 454
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011382575757575758,
+      "loss": 1.7848,
+      "step": 455
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011363636363636365,
+      "loss": 1.7756,
+      "step": 456
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001134469696969697,
+      "loss": 1.7964,
+      "step": 457
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011325757575757576,
+      "loss": 1.7604,
+      "step": 458
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011306818181818182,
+      "loss": 1.7914,
+      "step": 459
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001128787878787879,
+      "loss": 1.8059,
+      "step": 460
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011268939393939395,
+      "loss": 1.7647,
+      "step": 461
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011250000000000001,
+      "loss": 1.7526,
+      "step": 462
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011231060606060607,
+      "loss": 1.7736,
+      "step": 463
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011212121212121212,
+      "loss": 1.7449,
+      "step": 464
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011193181818181819,
+      "loss": 1.7636,
+      "step": 465
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011174242424242425,
+      "loss": 1.7846,
+      "step": 466
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001115530303030303,
+      "loss": 1.78,
+      "step": 467
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011136363636363636,
+      "loss": 1.7828,
+      "step": 468
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011117424242424244,
+      "loss": 1.729,
+      "step": 469
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001109848484848485,
+      "loss": 1.7145,
+      "step": 470
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011079545454545455,
+      "loss": 1.7189,
+      "step": 471
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011060606060606061,
+      "loss": 1.7628,
+      "step": 472
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011041666666666668,
+      "loss": 1.7399,
+      "step": 473
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011022727272727273,
+      "loss": 1.7561,
+      "step": 474
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011003787878787879,
+      "loss": 1.7979,
+      "step": 475
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010984848484848484,
+      "loss": 1.7673,
+      "step": 476
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010965909090909093,
+      "loss": 1.777,
+      "step": 477
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010946969696969698,
+      "loss": 1.7042,
+      "step": 478
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010928030303030304,
+      "loss": 1.7764,
+      "step": 479
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010909090909090909,
+      "loss": 1.6993,
+      "step": 480
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010890151515151516,
+      "loss": 1.7688,
+      "step": 481
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010871212121212122,
+      "loss": 1.7428,
+      "step": 482
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010852272727272727,
+      "loss": 1.675,
+      "step": 483
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010833333333333333,
+      "loss": 1.7183,
+      "step": 484
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010814393939393941,
+      "loss": 1.7305,
+      "step": 485
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010795454545454547,
+      "loss": 1.7541,
+      "step": 486
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010776515151515152,
+      "loss": 1.7074,
+      "step": 487
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010757575757575758,
+      "loss": 1.7093,
+      "step": 488
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010738636363636365,
+      "loss": 1.7354,
+      "step": 489
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001071969696969697,
+      "loss": 1.7415,
+      "step": 490
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010700757575757576,
+      "loss": 1.72,
+      "step": 491
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010681818181818181,
+      "loss": 1.7453,
+      "step": 492
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001066287878787879,
+      "loss": 1.7077,
+      "step": 493
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010643939393939395,
+      "loss": 1.6936,
+      "step": 494
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010625000000000001,
+      "loss": 1.7616,
+      "step": 495
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010606060606060606,
+      "loss": 1.7749,
+      "step": 496
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010587121212121212,
+      "loss": 1.7375,
+      "step": 497
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010568181818181819,
+      "loss": 1.7203,
+      "step": 498
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010549242424242424,
+      "loss": 1.7148,
+      "step": 499
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001053030303030303,
+      "loss": 1.7859,
+      "step": 500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010511363636363635,
+      "loss": 1.7478,
+      "step": 501
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010492424242424244,
+      "loss": 1.7091,
+      "step": 502
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010473484848484849,
+      "loss": 1.7112,
+      "step": 503
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010454545454545455,
+      "loss": 1.6967,
+      "step": 504
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001043560606060606,
+      "loss": 1.7431,
+      "step": 505
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010416666666666667,
+      "loss": 1.7065,
+      "step": 506
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010397727272727273,
+      "loss": 1.6955,
+      "step": 507
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010378787878787878,
+      "loss": 1.7375,
+      "step": 508
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010359848484848484,
+      "loss": 1.7056,
+      "step": 509
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010340909090909092,
+      "loss": 1.7044,
+      "step": 510
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010321969696969698,
+      "loss": 1.7204,
+      "step": 511
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010303030303030303,
+      "loss": 1.6801,
+      "step": 512
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010284090909090909,
+      "loss": 1.7381,
+      "step": 513
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010265151515151516,
+      "loss": 1.7064,
+      "step": 514
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010246212121212121,
+      "loss": 1.6973,
+      "step": 515
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010227272727272727,
+      "loss": 1.7295,
+      "step": 516
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010208333333333333,
+      "loss": 1.6991,
+      "step": 517
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010189393939393941,
+      "loss": 1.6986,
+      "step": 518
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010170454545454546,
+      "loss": 1.6989,
+      "step": 519
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010151515151515152,
+      "loss": 1.7009,
+      "step": 520
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010132575757575757,
+      "loss": 1.6919,
+      "step": 521
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010113636363636366,
+      "loss": 1.6955,
+      "step": 522
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010094696969696971,
+      "loss": 1.7177,
+      "step": 523
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010075757575757576,
+      "loss": 1.715,
+      "step": 524
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010056818181818181,
+      "loss": 1.6686,
+      "step": 525
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001003787878787879,
+      "loss": 1.771,
+      "step": 526
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010018939393939395,
+      "loss": 1.7024,
+      "step": 527
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001,
+      "loss": 1.7016,
+      "step": 528
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.981060606060606e-05,
+      "loss": 1.6501,
+      "step": 529
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.962121212121213e-05,
+      "loss": 1.6903,
+      "step": 530
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.943181818181819e-05,
+      "loss": 1.6806,
+      "step": 531
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.924242424242425e-05,
+      "loss": 1.7096,
+      "step": 532
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.90530303030303e-05,
+      "loss": 1.7307,
+      "step": 533
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.886363636363637e-05,
+      "loss": 1.6871,
+      "step": 534
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.867424242424242e-05,
+      "loss": 1.7457,
+      "step": 535
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.848484848484849e-05,
+      "loss": 1.6867,
+      "step": 536
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.829545454545455e-05,
+      "loss": 1.6789,
+      "step": 537
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.810606060606061e-05,
+      "loss": 1.6403,
+      "step": 538
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.791666666666667e-05,
+      "loss": 1.6697,
+      "step": 539
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.772727272727274e-05,
+      "loss": 1.7293,
+      "step": 540
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.75378787878788e-05,
+      "loss": 1.6998,
+      "step": 541
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.734848484848485e-05,
+      "loss": 1.693,
+      "step": 542
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.71590909090909e-05,
+      "loss": 1.664,
+      "step": 543
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.696969696969698e-05,
+      "loss": 1.7061,
+      "step": 544
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.678030303030303e-05,
+      "loss": 1.6631,
+      "step": 545
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.65909090909091e-05,
+      "loss": 1.6343,
+      "step": 546
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.640151515151516e-05,
+      "loss": 1.6939,
+      "step": 547
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.621212121212123e-05,
+      "loss": 1.669,
+      "step": 548
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.602272727272728e-05,
+      "loss": 1.6561,
+      "step": 549
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.583333333333334e-05,
+      "loss": 1.6675,
+      "step": 550
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.564393939393939e-05,
+      "loss": 1.7109,
+      "step": 551
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.545454545454546e-05,
+      "loss": 1.693,
+      "step": 552
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.526515151515152e-05,
+      "loss": 1.6557,
+      "step": 553
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.507575757575759e-05,
+      "loss": 1.6642,
+      "step": 554
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.488636363636364e-05,
+      "loss": 1.6674,
+      "step": 555
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.469696969696971e-05,
+      "loss": 1.6492,
+      "step": 556
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.450757575757577e-05,
+      "loss": 1.6915,
+      "step": 557
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.431818181818182e-05,
+      "loss": 1.7028,
+      "step": 558
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.412878787878788e-05,
+      "loss": 1.6749,
+      "step": 559
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.393939393939395e-05,
+      "loss": 1.6526,
+      "step": 560
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.375e-05,
+      "loss": 1.687,
+      "step": 561
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.356060606060606e-05,
+      "loss": 1.6632,
+      "step": 562
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.337121212121213e-05,
+      "loss": 1.7074,
+      "step": 563
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.318181818181818e-05,
+      "loss": 1.6164,
+      "step": 564
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.299242424242425e-05,
+      "loss": 1.6594,
+      "step": 565
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.280303030303031e-05,
+      "loss": 1.6603,
+      "step": 566
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.261363636363636e-05,
+      "loss": 1.6213,
+      "step": 567
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.242424242424242e-05,
+      "loss": 1.6899,
+      "step": 568
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.223484848484849e-05,
+      "loss": 1.6619,
+      "step": 569
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.204545454545454e-05,
+      "loss": 1.7035,
+      "step": 570
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.185606060606061e-05,
+      "loss": 1.6408,
+      "step": 571
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.166666666666667e-05,
+      "loss": 1.6506,
+      "step": 572
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.147727272727274e-05,
+      "loss": 1.658,
+      "step": 573
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.128787878787879e-05,
+      "loss": 1.6005,
+      "step": 574
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.109848484848486e-05,
+      "loss": 1.6821,
+      "step": 575
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.090909090909092e-05,
+      "loss": 1.6858,
+      "step": 576
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.071969696969697e-05,
+      "loss": 1.6933,
+      "step": 577
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.053030303030303e-05,
+      "loss": 1.6757,
+      "step": 578
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.03409090909091e-05,
+      "loss": 1.6107,
+      "step": 579
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.015151515151515e-05,
+      "loss": 1.5751,
+      "step": 580
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.996212121212122e-05,
+      "loss": 1.6168,
+      "step": 581
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.977272727272728e-05,
+      "loss": 1.6213,
+      "step": 582
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.958333333333335e-05,
+      "loss": 1.6243,
+      "step": 583
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.93939393939394e-05,
+      "loss": 1.6249,
+      "step": 584
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.920454545454546e-05,
+      "loss": 1.6529,
+      "step": 585
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.901515151515151e-05,
+      "loss": 1.626,
+      "step": 586
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.882575757575758e-05,
+      "loss": 1.6616,
+      "step": 587
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.863636363636364e-05,
+      "loss": 1.6622,
+      "step": 588
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.844696969696971e-05,
+      "loss": 1.5927,
+      "step": 589
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.825757575757576e-05,
+      "loss": 1.6351,
+      "step": 590
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.806818181818183e-05,
+      "loss": 1.6213,
+      "step": 591
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.787878787878789e-05,
+      "loss": 1.635,
+      "step": 592
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.768939393939394e-05,
+      "loss": 1.6406,
+      "step": 593
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.75e-05,
+      "loss": 1.6387,
+      "step": 594
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.731060606060605e-05,
+      "loss": 1.602,
+      "step": 595
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.712121212121212e-05,
+      "loss": 1.601,
+      "step": 596
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.693181818181818e-05,
+      "loss": 1.5855,
+      "step": 597
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.674242424242425e-05,
+      "loss": 1.6236,
+      "step": 598
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.65530303030303e-05,
+      "loss": 1.5999,
+      "step": 599
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.636363636363637e-05,
+      "loss": 1.6093,
+      "step": 600
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.617424242424243e-05,
+      "loss": 1.6602,
+      "step": 601
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.598484848484848e-05,
+      "loss": 1.599,
+      "step": 602
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.579545454545454e-05,
+      "loss": 1.6056,
+      "step": 603
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.560606060606061e-05,
+      "loss": 1.6377,
+      "step": 604
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.541666666666666e-05,
+      "loss": 1.5769,
+      "step": 605
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.522727272727273e-05,
+      "loss": 1.6219,
+      "step": 606
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.503787878787879e-05,
+      "loss": 1.5917,
+      "step": 607
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.484848484848486e-05,
+      "loss": 1.6019,
+      "step": 608
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.465909090909091e-05,
+      "loss": 1.6316,
+      "step": 609
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.446969696969697e-05,
+      "loss": 1.6327,
+      "step": 610
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.428030303030303e-05,
+      "loss": 1.6023,
+      "step": 611
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.40909090909091e-05,
+      "loss": 1.6087,
+      "step": 612
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.390151515151515e-05,
+      "loss": 1.6245,
+      "step": 613
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.371212121212122e-05,
+      "loss": 1.5957,
+      "step": 614
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.352272727272727e-05,
+      "loss": 1.6196,
+      "step": 615
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.333333333333334e-05,
+      "loss": 1.6364,
+      "step": 616
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.31439393939394e-05,
+      "loss": 1.5977,
+      "step": 617
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.295454545454547e-05,
+      "loss": 1.6018,
+      "step": 618
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.276515151515152e-05,
+      "loss": 1.5973,
+      "step": 619
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.257575757575758e-05,
+      "loss": 1.6216,
+      "step": 620
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.238636363636364e-05,
+      "loss": 1.6422,
+      "step": 621
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.21969696969697e-05,
+      "loss": 1.6401,
+      "step": 622
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.200757575757576e-05,
+      "loss": 1.6446,
+      "step": 623
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.181818181818183e-05,
+      "loss": 1.5791,
+      "step": 624
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.162878787878789e-05,
+      "loss": 1.5953,
+      "step": 625
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.143939393939395e-05,
+      "loss": 1.5941,
+      "step": 626
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.125000000000001e-05,
+      "loss": 1.5784,
+      "step": 627
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.106060606060607e-05,
+      "loss": 1.6024,
+      "step": 628
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.087121212121212e-05,
+      "loss": 1.6295,
+      "step": 629
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.068181818181818e-05,
+      "loss": 1.5905,
+      "step": 630
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.049242424242425e-05,
+      "loss": 1.6073,
+      "step": 631
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.03030303030303e-05,
+      "loss": 1.6104,
+      "step": 632
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.011363636363637e-05,
+      "loss": 1.6134,
+      "step": 633
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.992424242424243e-05,
+      "loss": 1.6569,
+      "step": 634
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.97348484848485e-05,
+      "loss": 1.5493,
+      "step": 635
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.954545454545455e-05,
+      "loss": 1.5767,
+      "step": 636
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.93560606060606e-05,
+      "loss": 1.5692,
+      "step": 637
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.916666666666666e-05,
+      "loss": 1.6116,
+      "step": 638
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.897727272727273e-05,
+      "loss": 1.5684,
+      "step": 639
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.878787878787879e-05,
+      "loss": 1.6177,
+      "step": 640
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.859848484848486e-05,
+      "loss": 1.6151,
+      "step": 641
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.840909090909091e-05,
+      "loss": 1.6293,
+      "step": 642
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.821969696969698e-05,
+      "loss": 1.6298,
+      "step": 643
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.803030303030304e-05,
+      "loss": 1.6073,
+      "step": 644
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.784090909090909e-05,
+      "loss": 1.5328,
+      "step": 645
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.765151515151515e-05,
+      "loss": 1.5895,
+      "step": 646
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.746212121212122e-05,
+      "loss": 1.5728,
+      "step": 647
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.727272727272727e-05,
+      "loss": 1.5449,
+      "step": 648
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.708333333333334e-05,
+      "loss": 1.5731,
+      "step": 649
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.68939393939394e-05,
+      "loss": 1.627,
+      "step": 650
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.670454545454547e-05,
+      "loss": 1.6139,
+      "step": 651
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.651515151515152e-05,
+      "loss": 1.5613,
+      "step": 652
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.632575757575758e-05,
+      "loss": 1.5734,
+      "step": 653
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.613636363636363e-05,
+      "loss": 1.5537,
+      "step": 654
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.59469696969697e-05,
+      "loss": 1.5886,
+      "step": 655
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.575757575757576e-05,
+      "loss": 1.5504,
+      "step": 656
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.556818181818183e-05,
+      "loss": 1.5613,
+      "step": 657
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.537878787878788e-05,
+      "loss": 1.5877,
+      "step": 658
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.518939393939395e-05,
+      "loss": 1.605,
+      "step": 659
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 1.5403,
+      "step": 660
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.481060606060606e-05,
+      "loss": 1.6039,
+      "step": 661
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.462121212121213e-05,
+      "loss": 1.5708,
+      "step": 662
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.443181818181817e-05,
+      "loss": 1.5692,
+      "step": 663
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.424242424242424e-05,
+      "loss": 1.5084,
+      "step": 664
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.40530303030303e-05,
+      "loss": 1.5982,
+      "step": 665
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.386363636363637e-05,
+      "loss": 1.5881,
+      "step": 666
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.367424242424242e-05,
+      "loss": 1.5593,
+      "step": 667
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.348484848484849e-05,
+      "loss": 1.5871,
+      "step": 668
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.329545454545455e-05,
+      "loss": 1.6134,
+      "step": 669
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.310606060606062e-05,
+      "loss": 1.5516,
+      "step": 670
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.291666666666667e-05,
+      "loss": 1.5691,
+      "step": 671
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.272727272727273e-05,
+      "loss": 1.5801,
+      "step": 672
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.253787878787878e-05,
+      "loss": 1.5684,
+      "step": 673
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.234848484848485e-05,
+      "loss": 1.5591,
+      "step": 674
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.215909090909091e-05,
+      "loss": 1.5727,
+      "step": 675
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.196969696969698e-05,
+      "loss": 1.6081,
+      "step": 676
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.178030303030303e-05,
+      "loss": 1.5884,
+      "step": 677
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.15909090909091e-05,
+      "loss": 1.5638,
+      "step": 678
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.140151515151516e-05,
+      "loss": 1.5614,
+      "step": 679
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.121212121212121e-05,
+      "loss": 1.5543,
+      "step": 680
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.102272727272727e-05,
+      "loss": 1.5801,
+      "step": 681
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.083333333333334e-05,
+      "loss": 1.5458,
+      "step": 682
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.06439393939394e-05,
+      "loss": 1.5567,
+      "step": 683
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.045454545454546e-05,
+      "loss": 1.5567,
+      "step": 684
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.026515151515152e-05,
+      "loss": 1.5638,
+      "step": 685
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.007575757575759e-05,
+      "loss": 1.5431,
+      "step": 686
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.988636363636364e-05,
+      "loss": 1.5729,
+      "step": 687
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.96969696969697e-05,
+      "loss": 1.5235,
+      "step": 688
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.950757575757575e-05,
+      "loss": 1.5753,
+      "step": 689
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.931818181818182e-05,
+      "loss": 1.5319,
+      "step": 690
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.912878787878788e-05,
+      "loss": 1.5847,
+      "step": 691
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.893939393939395e-05,
+      "loss": 1.5533,
+      "step": 692
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.875e-05,
+      "loss": 1.5665,
+      "step": 693
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.856060606060606e-05,
+      "loss": 1.5913,
+      "step": 694
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.837121212121213e-05,
+      "loss": 1.6011,
+      "step": 695
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.818181818181818e-05,
+      "loss": 1.5201,
+      "step": 696
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.799242424242424e-05,
+      "loss": 1.57,
+      "step": 697
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.78030303030303e-05,
+      "loss": 1.5381,
+      "step": 698
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.761363636363636e-05,
+      "loss": 1.5681,
+      "step": 699
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.742424242424242e-05,
+      "loss": 1.5542,
+      "step": 700
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.723484848484849e-05,
+      "loss": 1.5779,
+      "step": 701
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.704545454545455e-05,
+      "loss": 1.578,
+      "step": 702
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.685606060606061e-05,
+      "loss": 1.6131,
+      "step": 703
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 1.5085,
+      "step": 704
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.647727272727274e-05,
+      "loss": 1.4876,
+      "step": 705
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.628787878787878e-05,
+      "loss": 1.5071,
+      "step": 706
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.609848484848485e-05,
+      "loss": 1.574,
+      "step": 707
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.59090909090909e-05,
+      "loss": 1.5214,
+      "step": 708
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.571969696969697e-05,
+      "loss": 1.5382,
+      "step": 709
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.553030303030303e-05,
+      "loss": 1.5136,
+      "step": 710
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.53409090909091e-05,
+      "loss": 1.4807,
+      "step": 711
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.515151515151516e-05,
+      "loss": 1.491,
+      "step": 712
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.496212121212122e-05,
+      "loss": 1.5595,
+      "step": 713
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.477272727272728e-05,
+      "loss": 1.5342,
+      "step": 714
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.458333333333334e-05,
+      "loss": 1.5173,
+      "step": 715
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.439393939393939e-05,
+      "loss": 1.5353,
+      "step": 716
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.420454545454546e-05,
+      "loss": 1.4826,
+      "step": 717
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.401515151515152e-05,
+      "loss": 1.5404,
+      "step": 718
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.382575757575759e-05,
+      "loss": 1.5612,
+      "step": 719
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.363636363636364e-05,
+      "loss": 1.5203,
+      "step": 720
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.344696969696971e-05,
+      "loss": 1.52,
+      "step": 721
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.325757575757577e-05,
+      "loss": 1.5417,
+      "step": 722
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.306818181818182e-05,
+      "loss": 1.5352,
+      "step": 723
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.287878787878788e-05,
+      "loss": 1.4671,
+      "step": 724
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.268939393939395e-05,
+      "loss": 1.5739,
+      "step": 725
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.25e-05,
+      "loss": 1.4987,
+      "step": 726
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.231060606060606e-05,
+      "loss": 1.5145,
+      "step": 727
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.212121212121213e-05,
+      "loss": 1.5686,
+      "step": 728
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.193181818181818e-05,
+      "loss": 1.4872,
+      "step": 729
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.174242424242425e-05,
+      "loss": 1.4831,
+      "step": 730
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.15530303030303e-05,
+      "loss": 1.5242,
+      "step": 731
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.136363636363636e-05,
+      "loss": 1.5298,
+      "step": 732
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.117424242424242e-05,
+      "loss": 1.4941,
+      "step": 733
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.098484848484849e-05,
+      "loss": 1.5022,
+      "step": 734
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.079545454545454e-05,
+      "loss": 1.4947,
+      "step": 735
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.060606060606061e-05,
+      "loss": 1.4922,
+      "step": 736
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.041666666666667e-05,
+      "loss": 1.4796,
+      "step": 737
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.022727272727273e-05,
+      "loss": 1.4619,
+      "step": 738
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.0037878787878785e-05,
+      "loss": 1.5346,
+      "step": 739
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.9848484848484854e-05,
+      "loss": 1.4987,
+      "step": 740
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.965909090909091e-05,
+      "loss": 1.516,
+      "step": 741
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.946969696969697e-05,
+      "loss": 1.5075,
+      "step": 742
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.928030303030303e-05,
+      "loss": 1.4584,
+      "step": 743
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.90909090909091e-05,
+      "loss": 1.5285,
+      "step": 744
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.890151515151515e-05,
+      "loss": 1.5339,
+      "step": 745
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.871212121212122e-05,
+      "loss": 1.4811,
+      "step": 746
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.852272727272727e-05,
+      "loss": 1.5158,
+      "step": 747
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.833333333333334e-05,
+      "loss": 1.5523,
+      "step": 748
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.8143939393939395e-05,
+      "loss": 1.4911,
+      "step": 749
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.7954545454545464e-05,
+      "loss": 1.508,
+      "step": 750
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.776515151515152e-05,
+      "loss": 1.5273,
+      "step": 751
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.757575757575758e-05,
+      "loss": 1.5231,
+      "step": 752
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.738636363636364e-05,
+      "loss": 1.5269,
+      "step": 753
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.719696969696971e-05,
+      "loss": 1.5306,
+      "step": 754
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.700757575757576e-05,
+      "loss": 1.4519,
+      "step": 755
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.6818181818181825e-05,
+      "loss": 1.4976,
+      "step": 756
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.662878787878788e-05,
+      "loss": 1.5136,
+      "step": 757
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.643939393939395e-05,
+      "loss": 1.536,
+      "step": 758
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.6250000000000005e-05,
+      "loss": 1.4927,
+      "step": 759
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.606060606060606e-05,
+      "loss": 1.4931,
+      "step": 760
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.587121212121212e-05,
+      "loss": 1.4939,
+      "step": 761
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.568181818181818e-05,
+      "loss": 1.4446,
+      "step": 762
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.549242424242425e-05,
+      "loss": 1.5113,
+      "step": 763
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.5303030303030304e-05,
+      "loss": 1.4886,
+      "step": 764
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.5113636363636366e-05,
+      "loss": 1.5291,
+      "step": 765
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.492424242424242e-05,
+      "loss": 1.5204,
+      "step": 766
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.473484848484849e-05,
+      "loss": 1.4677,
+      "step": 767
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4545454545454546e-05,
+      "loss": 1.4499,
+      "step": 768
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.435606060606061e-05,
+      "loss": 1.4735,
+      "step": 769
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4166666666666664e-05,
+      "loss": 1.4688,
+      "step": 770
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.397727272727273e-05,
+      "loss": 1.4728,
+      "step": 771
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.378787878787879e-05,
+      "loss": 1.4791,
+      "step": 772
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.359848484848485e-05,
+      "loss": 1.5265,
+      "step": 773
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.340909090909091e-05,
+      "loss": 1.4775,
+      "step": 774
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.3219696969696976e-05,
+      "loss": 1.4531,
+      "step": 775
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.303030303030303e-05,
+      "loss": 1.4891,
+      "step": 776
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.2840909090909094e-05,
+      "loss": 1.4764,
+      "step": 777
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.265151515151515e-05,
+      "loss": 1.4864,
+      "step": 778
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.246212121212122e-05,
+      "loss": 1.521,
+      "step": 779
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.2272727272727274e-05,
+      "loss": 1.473,
+      "step": 780
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.208333333333334e-05,
+      "loss": 1.477,
+      "step": 781
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.189393939393939e-05,
+      "loss": 1.482,
+      "step": 782
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.170454545454546e-05,
+      "loss": 1.5387,
+      "step": 783
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.151515151515152e-05,
+      "loss": 1.4354,
+      "step": 784
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.132575757575758e-05,
+      "loss": 1.4792,
+      "step": 785
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.1136363636363635e-05,
+      "loss": 1.4589,
+      "step": 786
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.0946969696969704e-05,
+      "loss": 1.4883,
+      "step": 787
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.075757575757576e-05,
+      "loss": 1.4442,
+      "step": 788
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.056818181818183e-05,
+      "loss": 1.4387,
+      "step": 789
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.037878787878788e-05,
+      "loss": 1.5203,
+      "step": 790
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.018939393939395e-05,
+      "loss": 1.4634,
+      "step": 791
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5e-05,
+      "loss": 1.4734,
+      "step": 792
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.9810606060606065e-05,
+      "loss": 1.5079,
+      "step": 793
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.962121212121213e-05,
+      "loss": 1.4986,
+      "step": 794
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.943181818181818e-05,
+      "loss": 1.4132,
+      "step": 795
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.9242424242424245e-05,
+      "loss": 1.4992,
+      "step": 796
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.905303030303031e-05,
+      "loss": 1.4601,
+      "step": 797
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.886363636363637e-05,
+      "loss": 1.47,
+      "step": 798
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8674242424242425e-05,
+      "loss": 1.4575,
+      "step": 799
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.848484848484849e-05,
+      "loss": 1.4508,
+      "step": 800
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.829545454545455e-05,
+      "loss": 1.4806,
+      "step": 801
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.810606060606061e-05,
+      "loss": 1.475,
+      "step": 802
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.791666666666667e-05,
+      "loss": 1.4219,
+      "step": 803
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.772727272727273e-05,
+      "loss": 1.4651,
+      "step": 804
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.753787878787879e-05,
+      "loss": 1.519,
+      "step": 805
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.7348484848484855e-05,
+      "loss": 1.4887,
+      "step": 806
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.715909090909091e-05,
+      "loss": 1.4713,
+      "step": 807
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.696969696969697e-05,
+      "loss": 1.4588,
+      "step": 808
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.678030303030303e-05,
+      "loss": 1.4752,
+      "step": 809
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.659090909090909e-05,
+      "loss": 1.4722,
+      "step": 810
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6401515151515154e-05,
+      "loss": 1.4735,
+      "step": 811
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.621212121212121e-05,
+      "loss": 1.4893,
+      "step": 812
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.602272727272727e-05,
+      "loss": 1.4793,
+      "step": 813
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.5833333333333334e-05,
+      "loss": 1.4737,
+      "step": 814
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.5643939393939396e-05,
+      "loss": 1.5194,
+      "step": 815
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.545454545454546e-05,
+      "loss": 1.4461,
+      "step": 816
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.5265151515151514e-05,
+      "loss": 1.467,
+      "step": 817
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.5075757575757577e-05,
+      "loss": 1.5031,
+      "step": 818
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.488636363636364e-05,
+      "loss": 1.5085,
+      "step": 819
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.46969696969697e-05,
+      "loss": 1.4356,
+      "step": 820
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.450757575757576e-05,
+      "loss": 1.4645,
+      "step": 821
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.431818181818182e-05,
+      "loss": 1.4224,
+      "step": 822
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.412878787878788e-05,
+      "loss": 1.4675,
+      "step": 823
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.3939393939393944e-05,
+      "loss": 1.4892,
+      "step": 824
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.375e-05,
+      "loss": 1.5441,
+      "step": 825
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.356060606060606e-05,
+      "loss": 1.4205,
+      "step": 826
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.3371212121212124e-05,
+      "loss": 1.4849,
+      "step": 827
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.318181818181819e-05,
+      "loss": 1.4789,
+      "step": 828
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.299242424242424e-05,
+      "loss": 1.445,
+      "step": 829
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.2803030303030305e-05,
+      "loss": 1.4528,
+      "step": 830
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.261363636363637e-05,
+      "loss": 1.4588,
+      "step": 831
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.242424242424243e-05,
+      "loss": 1.4158,
+      "step": 832
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.2234848484848485e-05,
+      "loss": 1.4933,
+      "step": 833
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.204545454545455e-05,
+      "loss": 1.4294,
+      "step": 834
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.185606060606061e-05,
+      "loss": 1.4764,
+      "step": 835
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 1.4262,
+      "step": 836
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.1477272727272734e-05,
+      "loss": 1.3994,
+      "step": 837
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.128787878787879e-05,
+      "loss": 1.4912,
+      "step": 838
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.109848484848485e-05,
+      "loss": 1.4228,
+      "step": 839
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.0909090909090915e-05,
+      "loss": 1.4403,
+      "step": 840
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.071969696969698e-05,
+      "loss": 1.4738,
+      "step": 841
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.053030303030303e-05,
+      "loss": 1.4715,
+      "step": 842
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.034090909090909e-05,
+      "loss": 1.4354,
+      "step": 843
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.015151515151515e-05,
+      "loss": 1.4296,
+      "step": 844
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.996212121212121e-05,
+      "loss": 1.4773,
+      "step": 845
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.9772727272727275e-05,
+      "loss": 1.4813,
+      "step": 846
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.958333333333333e-05,
+      "loss": 1.4763,
+      "step": 847
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.939393939393939e-05,
+      "loss": 1.472,
+      "step": 848
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9204545454545456e-05,
+      "loss": 1.3975,
+      "step": 849
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.901515151515152e-05,
+      "loss": 1.4507,
+      "step": 850
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.8825757575757574e-05,
+      "loss": 1.4911,
+      "step": 851
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8636363636363636e-05,
+      "loss": 1.3785,
+      "step": 852
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.84469696969697e-05,
+      "loss": 1.3964,
+      "step": 853
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.825757575757576e-05,
+      "loss": 1.471,
+      "step": 854
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.8068181818181816e-05,
+      "loss": 1.4034,
+      "step": 855
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.787878787878788e-05,
+      "loss": 1.5032,
+      "step": 856
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.768939393939394e-05,
+      "loss": 1.488,
+      "step": 857
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 1.4341,
+      "step": 858
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7310606060606066e-05,
+      "loss": 1.4728,
+      "step": 859
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.712121212121212e-05,
+      "loss": 1.4146,
+      "step": 860
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.6931818181818184e-05,
+      "loss": 1.4241,
+      "step": 861
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6742424242424246e-05,
+      "loss": 1.4811,
+      "step": 862
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.655303030303031e-05,
+      "loss": 1.5406,
+      "step": 863
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6363636363636364e-05,
+      "loss": 1.4578,
+      "step": 864
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6174242424242427e-05,
+      "loss": 1.4487,
+      "step": 865
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.598484848484849e-05,
+      "loss": 1.4609,
+      "step": 866
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.579545454545455e-05,
+      "loss": 1.3872,
+      "step": 867
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.560606060606061e-05,
+      "loss": 1.4584,
+      "step": 868
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.541666666666667e-05,
+      "loss": 1.4516,
+      "step": 869
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.522727272727273e-05,
+      "loss": 1.4802,
+      "step": 870
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5037878787878794e-05,
+      "loss": 1.4247,
+      "step": 871
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.484848484848485e-05,
+      "loss": 1.4203,
+      "step": 872
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.465909090909091e-05,
+      "loss": 1.4793,
+      "step": 873
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4469696969696974e-05,
+      "loss": 1.4442,
+      "step": 874
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.428030303030303e-05,
+      "loss": 1.4779,
+      "step": 875
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.409090909090909e-05,
+      "loss": 1.4143,
+      "step": 876
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.390151515151515e-05,
+      "loss": 1.4758,
+      "step": 877
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.371212121212121e-05,
+      "loss": 1.4327,
+      "step": 878
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.352272727272727e-05,
+      "loss": 1.466,
+      "step": 879
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 1.498,
+      "step": 880
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.314393939393939e-05,
+      "loss": 1.4679,
+      "step": 881
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.295454545454545e-05,
+      "loss": 1.4606,
+      "step": 882
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.2765151515151515e-05,
+      "loss": 1.4416,
+      "step": 883
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.257575757575758e-05,
+      "loss": 1.4284,
+      "step": 884
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.238636363636364e-05,
+      "loss": 1.4554,
+      "step": 885
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2196969696969696e-05,
+      "loss": 1.4306,
+      "step": 886
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.200757575757576e-05,
+      "loss": 1.4751,
+      "step": 887
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.181818181818182e-05,
+      "loss": 1.4787,
+      "step": 888
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.162878787878788e-05,
+      "loss": 1.4345,
+      "step": 889
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.143939393939394e-05,
+      "loss": 1.4951,
+      "step": 890
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.125e-05,
+      "loss": 1.4212,
+      "step": 891
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.106060606060606e-05,
+      "loss": 1.4258,
+      "step": 892
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.0871212121212125e-05,
+      "loss": 1.4454,
+      "step": 893
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.068181818181818e-05,
+      "loss": 1.4717,
+      "step": 894
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0492424242424243e-05,
+      "loss": 1.45,
+      "step": 895
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0303030303030306e-05,
+      "loss": 1.4785,
+      "step": 896
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0113636363636365e-05,
+      "loss": 1.3477,
+      "step": 897
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.9924242424242427e-05,
+      "loss": 1.4807,
+      "step": 898
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.9734848484848486e-05,
+      "loss": 1.4737,
+      "step": 899
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.954545454545455e-05,
+      "loss": 1.4427,
+      "step": 900
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.935606060606061e-05,
+      "loss": 1.4793,
+      "step": 901
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.916666666666667e-05,
+      "loss": 1.4168,
+      "step": 902
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.8977272727272732e-05,
+      "loss": 1.4452,
+      "step": 903
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.878787878787879e-05,
+      "loss": 1.37,
+      "step": 904
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8598484848484853e-05,
+      "loss": 1.4097,
+      "step": 905
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8409090909090912e-05,
+      "loss": 1.3963,
+      "step": 906
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8219696969696975e-05,
+      "loss": 1.4226,
+      "step": 907
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.803030303030303e-05,
+      "loss": 1.4734,
+      "step": 908
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.784090909090909e-05,
+      "loss": 1.4221,
+      "step": 909
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7651515151515152e-05,
+      "loss": 1.4383,
+      "step": 910
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.746212121212121e-05,
+      "loss": 1.4968,
+      "step": 911
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7272727272727273e-05,
+      "loss": 1.4475,
+      "step": 912
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7083333333333332e-05,
+      "loss": 1.4358,
+      "step": 913
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.6893939393939394e-05,
+      "loss": 1.4536,
+      "step": 914
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6704545454545453e-05,
+      "loss": 1.419,
+      "step": 915
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6515151515151516e-05,
+      "loss": 1.416,
+      "step": 916
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6325757575757575e-05,
+      "loss": 1.4109,
+      "step": 917
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6136363636363637e-05,
+      "loss": 1.3907,
+      "step": 918
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5946969696969696e-05,
+      "loss": 1.4316,
+      "step": 919
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.575757575757576e-05,
+      "loss": 1.4484,
+      "step": 920
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5568181818181817e-05,
+      "loss": 1.5423,
+      "step": 921
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.537878787878788e-05,
+      "loss": 1.5007,
+      "step": 922
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.518939393939394e-05,
+      "loss": 1.4497,
+      "step": 923
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5e-05,
+      "loss": 1.4404,
+      "step": 924
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4810606060606064e-05,
+      "loss": 1.476,
+      "step": 925
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4621212121212123e-05,
+      "loss": 1.4526,
+      "step": 926
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4431818181818185e-05,
+      "loss": 1.4112,
+      "step": 927
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4242424242424244e-05,
+      "loss": 1.465,
+      "step": 928
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4053030303030306e-05,
+      "loss": 1.4395,
+      "step": 929
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3863636363636365e-05,
+      "loss": 1.4097,
+      "step": 930
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3674242424242428e-05,
+      "loss": 1.4487,
+      "step": 931
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3484848484848487e-05,
+      "loss": 1.4614,
+      "step": 932
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3295454545454546e-05,
+      "loss": 1.4373,
+      "step": 933
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3106060606060605e-05,
+      "loss": 1.4899,
+      "step": 934
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.2916666666666667e-05,
+      "loss": 1.4455,
+      "step": 935
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.272727272727273e-05,
+      "loss": 1.3954,
+      "step": 936
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2537878787878788e-05,
+      "loss": 1.456,
+      "step": 937
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.234848484848485e-05,
+      "loss": 1.4317,
+      "step": 938
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.215909090909091e-05,
+      "loss": 1.3824,
+      "step": 939
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1969696969696972e-05,
+      "loss": 1.4068,
+      "step": 940
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.178030303030303e-05,
+      "loss": 1.3588,
+      "step": 941
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1590909090909093e-05,
+      "loss": 1.405,
+      "step": 942
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1401515151515152e-05,
+      "loss": 1.4763,
+      "step": 943
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1212121212121215e-05,
+      "loss": 1.4424,
+      "step": 944
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1022727272727274e-05,
+      "loss": 1.4339,
+      "step": 945
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.0833333333333336e-05,
+      "loss": 1.4734,
+      "step": 946
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0643939393939395e-05,
+      "loss": 1.437,
+      "step": 947
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0454545454545457e-05,
+      "loss": 1.432,
+      "step": 948
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0265151515151516e-05,
+      "loss": 1.5139,
+      "step": 949
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0075757575757575e-05,
+      "loss": 1.4176,
+      "step": 950
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9886363636363638e-05,
+      "loss": 1.4205,
+      "step": 951
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9696969696969697e-05,
+      "loss": 1.4537,
+      "step": 952
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.950757575757576e-05,
+      "loss": 1.447,
+      "step": 953
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9318181818181818e-05,
+      "loss": 1.4213,
+      "step": 954
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.912878787878788e-05,
+      "loss": 1.4057,
+      "step": 955
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.893939393939394e-05,
+      "loss": 1.3947,
+      "step": 956
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 1.4504,
+      "step": 957
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.856060606060606e-05,
+      "loss": 1.4295,
+      "step": 958
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8371212121212123e-05,
+      "loss": 1.3945,
+      "step": 959
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 1.4175,
+      "step": 960
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7992424242424244e-05,
+      "loss": 1.4521,
+      "step": 961
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7803030303030303e-05,
+      "loss": 1.4259,
+      "step": 962
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7613636363636366e-05,
+      "loss": 1.3455,
+      "step": 963
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7424242424242425e-05,
+      "loss": 1.4303,
+      "step": 964
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7234848484848487e-05,
+      "loss": 1.4553,
+      "step": 965
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7045454545454546e-05,
+      "loss": 1.4166,
+      "step": 966
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.6856060606060605e-05,
+      "loss": 1.4248,
+      "step": 967
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 1.4213,
+      "step": 968
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6477272727272726e-05,
+      "loss": 1.4287,
+      "step": 969
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.628787878787879e-05,
+      "loss": 1.4804,
+      "step": 970
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6098484848484848e-05,
+      "loss": 1.3863,
+      "step": 971
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.590909090909091e-05,
+      "loss": 1.4326,
+      "step": 972
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.571969696969697e-05,
+      "loss": 1.3857,
+      "step": 973
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.553030303030303e-05,
+      "loss": 1.3745,
+      "step": 974
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.534090909090909e-05,
+      "loss": 1.4398,
+      "step": 975
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5151515151515153e-05,
+      "loss": 1.4498,
+      "step": 976
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4962121212121214e-05,
+      "loss": 1.4087,
+      "step": 977
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4772727272727274e-05,
+      "loss": 1.4844,
+      "step": 978
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4583333333333335e-05,
+      "loss": 1.4476,
+      "step": 979
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4393939393939396e-05,
+      "loss": 1.407,
+      "step": 980
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4204545454545456e-05,
+      "loss": 1.4173,
+      "step": 981
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4015151515151515e-05,
+      "loss": 1.4286,
+      "step": 982
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3825757575757576e-05,
+      "loss": 1.3656,
+      "step": 983
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3636363636363637e-05,
+      "loss": 1.3885,
+      "step": 984
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3446969696969697e-05,
+      "loss": 1.4205,
+      "step": 985
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3257575757575758e-05,
+      "loss": 1.395,
+      "step": 986
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3068181818181819e-05,
+      "loss": 1.4598,
+      "step": 987
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.287878787878788e-05,
+      "loss": 1.4488,
+      "step": 988
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.268939393939394e-05,
+      "loss": 1.4015,
+      "step": 989
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.25e-05,
+      "loss": 1.4647,
+      "step": 990
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2310606060606061e-05,
+      "loss": 1.4058,
+      "step": 991
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2121212121212122e-05,
+      "loss": 1.3689,
+      "step": 992
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1931818181818183e-05,
+      "loss": 1.4467,
+      "step": 993
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1742424242424243e-05,
+      "loss": 1.4298,
+      "step": 994
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1553030303030302e-05,
+      "loss": 1.4265,
+      "step": 995
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1363636363636365e-05,
+      "loss": 1.4099,
+      "step": 996
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1174242424242425e-05,
+      "loss": 1.4161,
+      "step": 997
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.0984848484848486e-05,
+      "loss": 1.4087,
+      "step": 998
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.0795454545454547e-05,
+      "loss": 1.4275,
+      "step": 999
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0606060606060607e-05,
+      "loss": 1.3889,
+      "step": 1000
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1056,
+  "num_train_epochs": 3,
+  "save_steps": 100,
+  "total_flos": 2.5331819242832794e+18,
+  "trial_name": null,
+  "trial_params": null
+}