diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,22201 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999098205428804,
+  "global_step": 3696,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.801801801801802e-06,
+      "loss": 1.2549,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.603603603603604e-06,
+      "loss": 0.2184,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.405405405405406e-06,
+      "loss": 1.3005,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.207207207207208e-06,
+      "loss": 1.231,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.00900900900901e-06,
+      "loss": 1.2242,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0810810810810812e-05,
+      "loss": 1.3305,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2612612612612611e-05,
+      "loss": 1.2781,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4414414414414416e-05,
+      "loss": 1.1921,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6216216216216218e-05,
+      "loss": 0.2325,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.801801801801802e-05,
+      "loss": 1.1979,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.981981981981982e-05,
+      "loss": 1.1864,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1621621621621624e-05,
+      "loss": 1.1251,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3423423423423425e-05,
+      "loss": 1.1432,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5225225225225222e-05,
+      "loss": 1.0877,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.702702702702703e-05,
+      "loss": 1.1297,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.882882882882883e-05,
+      "loss": 1.097,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.063063063063063e-05,
+      "loss": 1.1179,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2432432432432436e-05,
+      "loss": 1.0644,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4234234234234234e-05,
+      "loss": 1.0469,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.603603603603604e-05,
+      "loss": 1.0348,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.783783783783784e-05,
+      "loss": 0.9811,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.963963963963964e-05,
+      "loss": 1.0172,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1441441441441444e-05,
+      "loss": 1.0346,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.324324324324325e-05,
+      "loss": 1.0235,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5045045045045046e-05,
+      "loss": 1.0161,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.684684684684685e-05,
+      "loss": 0.9974,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8648648648648654e-05,
+      "loss": 0.9808,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.0450450450450445e-05,
+      "loss": 1.0073,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2252252252252256e-05,
+      "loss": 0.9999,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.405405405405406e-05,
+      "loss": 1.0056,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.585585585585585e-05,
+      "loss": 1.0049,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.765765765765766e-05,
+      "loss": 0.9773,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9459459459459466e-05,
+      "loss": 1.0028,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.126126126126126e-05,
+      "loss": 0.9824,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.306306306306306e-05,
+      "loss": 0.9852,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.486486486486487e-05,
+      "loss": 0.9743,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 1.0021,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.846846846846847e-05,
+      "loss": 0.2607,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.027027027027028e-05,
+      "loss": 0.9486,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.207207207207208e-05,
+      "loss": 0.9637,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.387387387387387e-05,
+      "loss": 0.9766,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.567567567567568e-05,
+      "loss": 0.2231,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.747747747747748e-05,
+      "loss": 1.0088,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.927927927927928e-05,
+      "loss": 0.9808,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.108108108108109e-05,
+      "loss": 0.2455,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.288288288288289e-05,
+      "loss": 0.9537,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.468468468468469e-05,
+      "loss": 0.9446,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.64864864864865e-05,
+      "loss": 0.938,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.82882882882883e-05,
+      "loss": 0.9121,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.009009009009009e-05,
+      "loss": 0.9799,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.18918918918919e-05,
+      "loss": 0.9839,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.36936936936937e-05,
+      "loss": 0.9343,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.54954954954955e-05,
+      "loss": 0.9333,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.729729729729731e-05,
+      "loss": 0.9524,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.90990990990991e-05,
+      "loss": 0.9356,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010090090090090089,
+      "loss": 0.992,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001027027027027027,
+      "loss": 0.9581,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010450450450450451,
+      "loss": 0.9909,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010630630630630631,
+      "loss": 0.9428,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010810810810810812,
+      "loss": 0.9327,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010990990990990993,
+      "loss": 0.9592,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001117117117117117,
+      "loss": 0.9422,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011351351351351351,
+      "loss": 0.9401,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011531531531531532,
+      "loss": 0.9417,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011711711711711712,
+      "loss": 0.9448,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011891891891891893,
+      "loss": 0.2113,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012072072072072073,
+      "loss": 0.9931,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012252252252252253,
+      "loss": 0.2259,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012432432432432433,
+      "loss": 0.9143,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012612612612612612,
+      "loss": 0.9924,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012792792792792795,
+      "loss": 0.8992,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012972972972972974,
+      "loss": 0.9193,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013153153153153154,
+      "loss": 0.9116,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013333333333333334,
+      "loss": 0.8999,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013513513513513514,
+      "loss": 0.9441,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013693693693693693,
+      "loss": 0.9155,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013873873873873876,
+      "loss": 0.9076,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014054054054054056,
+      "loss": 0.9353,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014234234234234233,
+      "loss": 0.9202,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014414414414414415,
+      "loss": 0.9535,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014594594594594595,
+      "loss": 0.2384,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014774774774774775,
+      "loss": 0.9252,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014954954954954957,
+      "loss": 0.9362,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015135135135135137,
+      "loss": 0.9588,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015315315315315314,
+      "loss": 0.9265,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015495495495495496,
+      "loss": 0.9166,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015675675675675676,
+      "loss": 0.9028,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015855855855855856,
+      "loss": 0.8922,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016036036036036038,
+      "loss": 0.9108,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016216216216216218,
+      "loss": 0.9228,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016396396396396395,
+      "loss": 0.2055,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016576576576576578,
+      "loss": 0.9235,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016756756756756757,
+      "loss": 0.9267,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016936936936936937,
+      "loss": 0.9238,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001711711711711712,
+      "loss": 0.9426,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000172972972972973,
+      "loss": 0.9182,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017477477477477476,
+      "loss": 0.938,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001765765765765766,
+      "loss": 0.9052,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017837837837837839,
+      "loss": 0.949,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018018018018018018,
+      "loss": 0.9206,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000181981981981982,
+      "loss": 0.9297,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001837837837837838,
+      "loss": 0.1928,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018558558558558558,
+      "loss": 0.9019,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001873873873873874,
+      "loss": 0.9244,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001891891891891892,
+      "loss": 0.9446,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000190990990990991,
+      "loss": 0.9349,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019279279279279282,
+      "loss": 0.9403,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019459459459459462,
+      "loss": 0.9212,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001963963963963964,
+      "loss": 0.9176,
+      "step": 109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001981981981981982,
+      "loss": 0.8649,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.9541,
+      "step": 111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999996160352093,
+      "loss": 0.2239,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999984641411314,
+      "loss": 0.9042,
+      "step": 113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999996544318651,
+      "loss": 0.9123,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999938565692428,
+      "loss": 0.9439,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999904008949706,
+      "loss": 0.2189,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999861772984878,
+      "loss": 0.1975,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999811857830378,
+      "loss": 0.8993,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999754263524543,
+      "loss": 0.9447,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999688990111601,
+      "loss": 0.9241,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999961603764167,
+      "loss": 0.9382,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999953540617078,
+      "loss": 0.9146,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999447095760847,
+      "loss": 0.8826,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999351106479685,
+      "loss": 0.9054,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999247438401012,
+      "loss": 0.9087,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999136091604434,
+      "loss": 0.9231,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999901706617546,
+      "loss": 0.1973,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999889036220549,
+      "loss": 0.9514,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998755979791827,
+      "loss": 0.8639,
+      "step": 129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998613919037665,
+      "loss": 0.924,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199984641800521,
+      "loss": 0.9754,
+      "step": 131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999830676295012,
+      "loss": 0.8931,
+      "step": 132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199981416678526,
+      "loss": 0.8923,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997968894886338,
+      "loss": 0.9552,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997788444184002,
+      "loss": 0.9083,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997600315884166,
+      "loss": 0.9165,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997404510131302,
+      "loss": 0.9582,
+      "step": 137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997201027075772,
+      "loss": 0.911,
+      "step": 138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999698986687384,
+      "loss": 0.8876,
+      "step": 139
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996771029687658,
+      "loss": 0.9144,
+      "step": 140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996544515685281,
+      "loss": 0.8441,
+      "step": 141
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999631032504065,
+      "loss": 0.915,
+      "step": 142
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999606845793362,
+      "loss": 0.9146,
+      "step": 143
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995818914549912,
+      "loss": 0.8621,
+      "step": 144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999556169508117,
+      "loss": 0.9273,
+      "step": 145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995296799724913,
+      "loss": 0.2276,
+      "step": 146
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995024228684565,
+      "loss": 0.9171,
+      "step": 147
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999474398216944,
+      "loss": 0.8446,
+      "step": 148
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994456060394754,
+      "loss": 0.8931,
+      "step": 149
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199941604635816,
+      "loss": 0.8602,
+      "step": 150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999385719195698,
+      "loss": 0.8877,
+      "step": 151
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993546245753788,
+      "loss": 0.9155,
+      "step": 152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999322762521081,
+      "loss": 0.905,
+      "step": 153
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992901330572713,
+      "loss": 0.8671,
+      "step": 154
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992567362090083,
+      "loss": 0.8848,
+      "step": 155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992225720019376,
+      "loss": 0.9286,
+      "step": 156
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999187640462295,
+      "loss": 0.9241,
+      "step": 157
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991519416169052,
+      "loss": 0.9198,
+      "step": 158
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991154754931832,
+      "loss": 0.8727,
+      "step": 159
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990782421191317,
+      "loss": 0.9646,
+      "step": 160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990402415233436,
+      "loss": 0.8722,
+      "step": 161
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990014737350005,
+      "loss": 0.8856,
+      "step": 162
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989619387838733,
+      "loss": 0.2361,
+      "step": 163
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989216367003227,
+      "loss": 0.8889,
+      "step": 164
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001998880567515297,
+      "loss": 0.8544,
+      "step": 165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001998838731260335,
+      "loss": 0.9419,
+      "step": 166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987961279675637,
+      "loss": 0.9427,
+      "step": 167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987527576696996,
+      "loss": 0.8691,
+      "step": 168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998708620400048,
+      "loss": 0.8826,
+      "step": 169
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019986637161925033,
+      "loss": 0.893,
+      "step": 170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019986180450815485,
+      "loss": 0.8804,
+      "step": 171
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998571607102256,
+      "loss": 0.9048,
+      "step": 172
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998524402290287,
+      "loss": 0.8839,
+      "step": 173
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998476430681891,
+      "loss": 0.8877,
+      "step": 174
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019984276923139076,
+      "loss": 0.9343,
+      "step": 175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019983781872237634,
+      "loss": 0.8756,
+      "step": 176
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019983279154494758,
+      "loss": 0.9241,
+      "step": 177
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019982768770296496,
+      "loss": 0.2098,
+      "step": 178
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019982250720034785,
+      "loss": 0.8656,
+      "step": 179
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019981725004107454,
+      "loss": 0.8686,
+      "step": 180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019981191622918217,
+      "loss": 0.8586,
+      "step": 181
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019980650576876668,
+      "loss": 0.9273,
+      "step": 182
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019980101866398298,
+      "loss": 0.9018,
+      "step": 183
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019979545491904472,
+      "loss": 0.9232,
+      "step": 184
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019978981453822452,
+      "loss": 0.8932,
+      "step": 185
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019978409752585376,
+      "loss": 0.895,
+      "step": 186
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019977830388632275,
+      "loss": 0.9184,
+      "step": 187
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019977243362408054,
+      "loss": 0.8843,
+      "step": 188
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997664867436351,
+      "loss": 0.879,
+      "step": 189
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997604632495532,
+      "loss": 0.8602,
+      "step": 190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997543631464605,
+      "loss": 0.9594,
+      "step": 191
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019974818643904142,
+      "loss": 0.8552,
+      "step": 192
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019974193313203925,
+      "loss": 0.8447,
+      "step": 193
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019973560323025607,
+      "loss": 0.9258,
+      "step": 194
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019972919673855283,
+      "loss": 0.8995,
+      "step": 195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019972271366184923,
+      "loss": 0.9,
+      "step": 196
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019971615400512382,
+      "loss": 0.8784,
+      "step": 197
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019970951777341397,
+      "loss": 0.8761,
+      "step": 198
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019970280497181583,
+      "loss": 0.8682,
+      "step": 199
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001996960156054844,
+      "loss": 0.8836,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019968914967963337,
+      "loss": 0.9098,
+      "step": 201
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019968220719953532,
+      "loss": 0.9232,
+      "step": 202
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019967518817052158,
+      "loss": 0.8976,
+      "step": 203
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019966809259798227,
+      "loss": 0.8851,
+      "step": 204
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996609204873663,
+      "loss": 0.9086,
+      "step": 205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019965367184418136,
+      "loss": 0.2166,
+      "step": 206
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996463466739938,
+      "loss": 0.8756,
+      "step": 207
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019963894498242898,
+      "loss": 0.884,
+      "step": 208
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019963146677517082,
+      "loss": 0.9005,
+      "step": 209
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000199623912057962,
+      "loss": 0.8621,
+      "step": 210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019961628083660406,
+      "loss": 0.9099,
+      "step": 211
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019960857311695723,
+      "loss": 0.9014,
+      "step": 212
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019960078890494053,
+      "loss": 0.9101,
+      "step": 213
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019959292820653167,
+      "loss": 0.8628,
+      "step": 214
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019958499102776706,
+      "loss": 0.9077,
+      "step": 215
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019957697737474196,
+      "loss": 0.8952,
+      "step": 216
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019956888725361025,
+      "loss": 0.9146,
+      "step": 217
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995607206705846,
+      "loss": 0.8841,
+      "step": 218
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019955247763193637,
+      "loss": 0.9465,
+      "step": 219
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019954415814399563,
+      "loss": 0.862,
+      "step": 220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019953576221315116,
+      "loss": 0.8902,
+      "step": 221
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995272898458504,
+      "loss": 0.8526,
+      "step": 222
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995187410485996,
+      "loss": 0.862,
+      "step": 223
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019951011582796359,
+      "loss": 0.2277,
+      "step": 224
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019950141419056597,
+      "loss": 0.8794,
+      "step": 225
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019949263614308892,
+      "loss": 0.8917,
+      "step": 226
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019948378169227344,
+      "loss": 0.9463,
+      "step": 227
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994748508449191,
+      "loss": 0.8854,
+      "step": 228
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019946584360788414,
+      "loss": 0.9069,
+      "step": 229
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994567599880855,
+      "loss": 0.8819,
+      "step": 230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019944759999249872,
+      "loss": 0.8956,
+      "step": 231
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994383636281581,
+      "loss": 0.8797,
+      "step": 232
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019942905090215645,
+      "loss": 0.859,
+      "step": 233
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019941966182164533,
+      "loss": 0.9095,
+      "step": 234
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019941019639383491,
+      "loss": 0.8738,
+      "step": 235
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019940065462599392,
+      "loss": 0.8881,
+      "step": 236
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019939103652544984,
+      "loss": 0.8846,
+      "step": 237
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019938134209958861,
+      "loss": 0.9079,
+      "step": 238
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019937157135585488,
+      "loss": 0.8906,
+      "step": 239
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019936172430175196,
+      "loss": 0.8833,
+      "step": 240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019935180094484163,
+      "loss": 0.9215,
+      "step": 241
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019934180129274433,
+      "loss": 0.8703,
+      "step": 242
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019933172535313914,
+      "loss": 0.8853,
+      "step": 243
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019932157313376362,
+      "loss": 0.8612,
+      "step": 244
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019931134464241398,
+      "loss": 0.8853,
+      "step": 245
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000199301039886945,
+      "loss": 0.8704,
+      "step": 246
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019929065887526996,
+      "loss": 0.8452,
+      "step": 247
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019928020161536078,
+      "loss": 0.8665,
+      "step": 248
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992696681152479,
+      "loss": 0.8721,
+      "step": 249
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992590583830203,
+      "loss": 0.8685,
+      "step": 250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992483724268255,
+      "loss": 0.9141,
+      "step": 251
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019923761025486956,
+      "loss": 0.8788,
+      "step": 252
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992267718754171,
+      "loss": 0.9049,
+      "step": 253
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992158572967912,
+      "loss": 0.919,
+      "step": 254
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019920486652737348,
+      "loss": 0.9164,
+      "step": 255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019919379957560415,
+      "loss": 0.901,
+      "step": 256
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019918265644998173,
+      "loss": 0.9082,
+      "step": 257
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019917143715906345,
+      "loss": 0.8968,
+      "step": 258
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001991601417114649,
+      "loss": 0.9008,
+      "step": 259
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019914877011586017,
+      "loss": 0.9139,
+      "step": 260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001991373223809819,
+      "loss": 0.8845,
+      "step": 261
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019912579851562109,
+      "loss": 0.2111,
+      "step": 262
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001991141985286273,
+      "loss": 0.8992,
+      "step": 263
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001991025224289085,
+      "loss": 0.9037,
+      "step": 264
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990907702254311,
+      "loss": 0.9061,
+      "step": 265
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019907894192721992,
+      "loss": 0.8469,
+      "step": 266
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019906703754335832,
+      "loss": 0.9027,
+      "step": 267
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019905505708298798,
+      "loss": 0.8736,
+      "step": 268
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990430005553091,
+      "loss": 0.8673,
+      "step": 269
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990308679695802,
+      "loss": 0.8924,
+      "step": 270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019901865933511835,
+      "loss": 0.8356,
+      "step": 271
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019900637466129876,
+      "loss": 0.2139,
+      "step": 272
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019899401395755533,
+      "loss": 0.8951,
+      "step": 273
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019898157723338017,
+      "loss": 0.8715,
+      "step": 274
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019896906449832374,
+      "loss": 0.8511,
+      "step": 275
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019895647576199506,
+      "loss": 0.9202,
+      "step": 276
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001989438110340613,
+      "loss": 0.906,
+      "step": 277
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019893107032424812,
+      "loss": 0.9103,
+      "step": 278
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019891825364233947,
+      "loss": 0.9027,
+      "step": 279
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019890536099817767,
+      "loss": 0.8928,
+      "step": 280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019889239240166337,
+      "loss": 0.8833,
+      "step": 281
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019887934786275553,
+      "loss": 0.8899,
+      "step": 282
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001988662273914714,
+      "loss": 0.2347,
+      "step": 283
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019885303099788666,
+      "loss": 0.8586,
+      "step": 284
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019883975869213516,
+      "loss": 0.873,
+      "step": 285
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001988264104844091,
+      "loss": 0.861,
+      "step": 286
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019881298638495893,
+      "loss": 0.894,
+      "step": 287
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019879948640409348,
+      "loss": 0.8853,
+      "step": 288
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019878591055217973,
+      "loss": 0.9272,
+      "step": 289
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019877225883964305,
+      "loss": 0.9038,
+      "step": 290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019875853127696692,
+      "loss": 0.8901,
+      "step": 291
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019874472787469313,
+      "loss": 0.9171,
+      "step": 292
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987308486434218,
+      "loss": 0.8429,
+      "step": 293
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019871689359381116,
+      "loss": 0.8731,
+      "step": 294
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987028627365777,
+      "loss": 0.8839,
+      "step": 295
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019868875608249612,
+      "loss": 0.8805,
+      "step": 296
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001986745736423994,
+      "loss": 0.8635,
+      "step": 297
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019866031542717854,
+      "loss": 0.8914,
+      "step": 298
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019864598144778297,
+      "loss": 0.2388,
+      "step": 299
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001986315717152201,
+      "loss": 0.8814,
+      "step": 300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001986170862405556,
+      "loss": 0.8736,
+      "step": 301
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001986025250349133,
+      "loss": 0.8817,
+      "step": 302
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985878881094752,
+      "loss": 0.8491,
+      "step": 303
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019857317547548143,
+      "loss": 0.8683,
+      "step": 304
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985583871442302,
+      "loss": 0.8547,
+      "step": 305
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019854352312707798,
+      "loss": 0.8614,
+      "step": 306
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019852858343543924,
+      "loss": 0.8538,
+      "step": 307
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019851356808078664,
+      "loss": 0.8897,
+      "step": 308
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019849847707465088,
+      "loss": 0.8833,
+      "step": 309
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019848331042862083,
+      "loss": 0.866,
+      "step": 310
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019846806815434337,
+      "loss": 0.8787,
+      "step": 311
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001984527502635235,
+      "loss": 0.8812,
+      "step": 312
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019843735676792435,
+      "loss": 0.9121,
+      "step": 313
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019842188767936695,
+      "loss": 0.8659,
+      "step": 314
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001984063430097305,
+      "loss": 0.8679,
+      "step": 315
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019839072277095222,
+      "loss": 0.8891,
+      "step": 316
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019837502697502734,
+      "loss": 0.8569,
+      "step": 317
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019835925563400914,
+      "loss": 0.2594,
+      "step": 318
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001983434087600089,
+      "loss": 0.8487,
+      "step": 319
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001983274863651959,
+      "loss": 0.9074,
+      "step": 320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001983114884617974,
+      "loss": 0.8968,
+      "step": 321
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001982954150620987,
+      "loss": 0.8905,
+      "step": 322
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019827926617844302,
+      "loss": 0.903,
+      "step": 323
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019826304182323155,
+      "loss": 0.881,
+      "step": 324
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019824674200892345,
+      "loss": 0.9594,
+      "step": 325
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019823036674803586,
+      "loss": 0.8999,
+      "step": 326
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019821391605314383,
+      "loss": 0.9164,
+      "step": 327
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981973899368803,
+      "loss": 0.919,
+      "step": 328
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981807884119362,
+      "loss": 0.8625,
+      "step": 329
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019816411149106028,
+      "loss": 0.8037,
+      "step": 330
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981473591870593,
+      "loss": 0.8945,
+      "step": 331
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019813053151279783,
+      "loss": 0.8432,
+      "step": 332
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019811362848119832,
+      "loss": 0.9018,
+      "step": 333
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019809665010524113,
+      "loss": 0.8644,
+      "step": 334
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019807959639796442,
+      "loss": 0.8811,
+      "step": 335
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001980624673724643,
+      "loss": 0.942,
+      "step": 336
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001980452630418946,
+      "loss": 0.8584,
+      "step": 337
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019802798341946707,
+      "loss": 0.8651,
+      "step": 338
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019801062851845122,
+      "loss": 0.8618,
+      "step": 339
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019799319835217438,
+      "loss": 0.8805,
+      "step": 340
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019797569293402174,
+      "loss": 0.9169,
+      "step": 341
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019795811227743619,
+      "loss": 0.8908,
+      "step": 342
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019794045639591845,
+      "loss": 0.8731,
+      "step": 343
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019792272530302695,
+      "loss": 0.9042,
+      "step": 344
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000197904919012378,
+      "loss": 0.8996,
+      "step": 345
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019788703753764552,
+      "loss": 0.898,
+      "step": 346
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019786908089256127,
+      "loss": 0.8357,
+      "step": 347
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001978510490909146,
+      "loss": 0.9077,
+      "step": 348
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019783294214655278,
+      "loss": 0.863,
+      "step": 349
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019781476007338058,
+      "loss": 0.8701,
+      "step": 350
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019779650288536058,
+      "loss": 0.8561,
+      "step": 351
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019777817059651303,
+      "loss": 0.2591,
+      "step": 352
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019775976322091582,
+      "loss": 0.2644,
+      "step": 353
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019774128077270451,
+      "loss": 0.9134,
+      "step": 354
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019772272326607236,
+      "loss": 0.9329,
+      "step": 355
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001977040907152702,
+      "loss": 0.8756,
+      "step": 356
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019768538313460647,
+      "loss": 0.8937,
+      "step": 357
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019766660053844735,
+      "loss": 0.8724,
+      "step": 358
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019764774294121656,
+      "loss": 0.8652,
+      "step": 359
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019762881035739532,
+      "loss": 0.9094,
+      "step": 360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019760980280152256,
+      "loss": 0.8419,
+      "step": 361
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001975907202881948,
+      "loss": 0.8605,
+      "step": 362
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019757156283206598,
+      "loss": 0.8832,
+      "step": 363
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019755233044784775,
+      "loss": 0.9098,
+      "step": 364
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001975330231503092,
+      "loss": 0.8543,
+      "step": 365
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019751364095427692,
+      "loss": 0.8594,
+      "step": 366
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019749418387463518,
+      "loss": 0.9069,
+      "step": 367
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019747465192632556,
+      "loss": 0.95,
+      "step": 368
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019745504512434724,
+      "loss": 0.9162,
+      "step": 369
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001974353634837569,
+      "loss": 0.9082,
+      "step": 370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019741560701966863,
+      "loss": 0.8605,
+      "step": 371
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019739577574725395,
+      "loss": 0.8198,
+      "step": 372
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019737586968174197,
+      "loss": 0.8586,
+      "step": 373
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019735588883841912,
+      "loss": 0.9064,
+      "step": 374
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019733583323262925,
+      "loss": 0.871,
+      "step": 375
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001973157028797737,
+      "loss": 0.8744,
+      "step": 376
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019729549779531108,
+      "loss": 0.8783,
+      "step": 377
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019727521799475754,
+      "loss": 0.9238,
+      "step": 378
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019725486349368655,
+      "loss": 0.8907,
+      "step": 379
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001972344343077289,
+      "loss": 0.8596,
+      "step": 380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019721393045257277,
+      "loss": 0.8823,
+      "step": 381
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019719335194396366,
+      "loss": 0.8748,
+      "step": 382
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019717269879770448,
+      "loss": 0.8549,
+      "step": 383
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019715197102965535,
+      "loss": 0.89,
+      "step": 384
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001971311686557337,
+      "loss": 0.8543,
+      "step": 385
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019711029169191434,
+      "loss": 0.9152,
+      "step": 386
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001970893401542293,
+      "loss": 0.8503,
+      "step": 387
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019706831405876787,
+      "loss": 0.8918,
+      "step": 388
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019704721342167662,
+      "loss": 0.8517,
+      "step": 389
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019702603825915934,
+      "loss": 0.8511,
+      "step": 390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001970047885874771,
+      "loss": 0.9107,
+      "step": 391
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019698346442294812,
+      "loss": 0.8963,
+      "step": 392
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019696206578194786,
+      "loss": 0.8583,
+      "step": 393
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019694059268090896,
+      "loss": 0.8698,
+      "step": 394
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001969190451363213,
+      "loss": 0.861,
+      "step": 395
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019689742316473182,
+      "loss": 0.893,
+      "step": 396
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019687572678274467,
+      "loss": 0.8524,
+      "step": 397
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001968539560070212,
+      "loss": 0.8251,
+      "step": 398
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019683211085427975,
+      "loss": 0.8946,
+      "step": 399
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019681019134129595,
+      "loss": 0.8461,
+      "step": 400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019678819748490236,
+      "loss": 0.8673,
+      "step": 401
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001967661293019888,
+      "loss": 0.8921,
+      "step": 402
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019674398680950202,
+      "loss": 0.8652,
+      "step": 403
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019672177002444586,
+      "loss": 0.8579,
+      "step": 404
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019669947896388136,
+      "loss": 0.8815,
+      "step": 405
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019667711364492637,
+      "loss": 0.9109,
+      "step": 406
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019665467408475594,
+      "loss": 0.8374,
+      "step": 407
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019663216030060208,
+      "loss": 0.8596,
+      "step": 408
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019660957230975377,
+      "loss": 0.8959,
+      "step": 409
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019658691012955698,
+      "loss": 0.8857,
+      "step": 410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001965641737774147,
+      "loss": 0.8996,
+      "step": 411
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019654136327078682,
+      "loss": 0.349,
+      "step": 412
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019651847862719023,
+      "loss": 0.8312,
+      "step": 413
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001964955198641987,
+      "loss": 0.8843,
+      "step": 414
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019647248699944297,
+      "loss": 0.8928,
+      "step": 415
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019644938005061063,
+      "loss": 0.8992,
+      "step": 416
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019642619903544623,
+      "loss": 0.3051,
+      "step": 417
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019640294397175106,
+      "loss": 0.8986,
+      "step": 418
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001963796148773835,
+      "loss": 0.8848,
+      "step": 419
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019635621177025857,
+      "loss": 0.9051,
+      "step": 420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019633273466834825,
+      "loss": 0.2733,
+      "step": 421
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019630918358968128,
+      "loss": 0.8771,
+      "step": 422
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019628555855234322,
+      "loss": 0.9048,
+      "step": 423
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019626185957447647,
+      "loss": 0.881,
+      "step": 424
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001962380866742801,
+      "loss": 0.9096,
+      "step": 425
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019621423987001014,
+      "loss": 0.9012,
+      "step": 426
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019619031917997916,
+      "loss": 0.2559,
+      "step": 427
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001961663246225566,
+      "loss": 0.8941,
+      "step": 428
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001961422562161686,
+      "loss": 0.8976,
+      "step": 429
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019611811397929799,
+      "loss": 0.8871,
+      "step": 430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001960938979304843,
+      "loss": 0.8908,
+      "step": 431
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019606960808832375,
+      "loss": 0.911,
+      "step": 432
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019604524447146927,
+      "loss": 0.8676,
+      "step": 433
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019602080709863034,
+      "loss": 0.9192,
+      "step": 434
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019599629598857315,
+      "loss": 0.856,
+      "step": 435
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019597171116012056,
+      "loss": 0.8602,
+      "step": 436
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019594705263215197,
+      "loss": 0.9047,
+      "step": 437
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019592232042360333,
+      "loss": 0.918,
+      "step": 438
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001958975145534673,
+      "loss": 0.2546,
+      "step": 439
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019587263504079302,
+      "loss": 0.2523,
+      "step": 440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019584768190468625,
+      "loss": 0.8827,
+      "step": 441
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019582265516430915,
+      "loss": 0.8936,
+      "step": 442
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019579755483888056,
+      "loss": 0.846,
+      "step": 443
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019577238094767573,
+      "loss": 0.8216,
+      "step": 444
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019574713351002646,
+      "loss": 0.8346,
+      "step": 445
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000195721812545321,
+      "loss": 0.2442,
+      "step": 446
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019569641807300406,
+      "loss": 0.8727,
+      "step": 447
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001956709501125768,
+      "loss": 0.8568,
+      "step": 448
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019564540868359685,
+      "loss": 0.8545,
+      "step": 449
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001956197938056782,
+      "loss": 0.8739,
+      "step": 450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019559410549849126,
+      "loss": 0.8502,
+      "step": 451
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001955683437817629,
+      "loss": 0.85,
+      "step": 452
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019554250867527624,
+      "loss": 0.8798,
+      "step": 453
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019551660019887084,
+      "loss": 0.8958,
+      "step": 454
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001954906183724426,
+      "loss": 0.8494,
+      "step": 455
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019546456321594376,
+      "loss": 0.8757,
+      "step": 456
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019543843474938277,
+      "loss": 0.8776,
+      "step": 457
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019541223299282453,
+      "loss": 0.8141,
+      "step": 458
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019538595796639007,
+      "loss": 0.9098,
+      "step": 459
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019535960969025682,
+      "loss": 0.8558,
+      "step": 460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019533318818465838,
+      "loss": 0.867,
+      "step": 461
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001953066934698846,
+      "loss": 0.9045,
+      "step": 462
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019528012556628153,
+      "loss": 0.252,
+      "step": 463
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019525348449425154,
+      "loss": 0.8949,
+      "step": 464
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000195226770274253,
+      "loss": 0.8354,
+      "step": 465
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019519998292680062,
+      "loss": 0.8684,
+      "step": 466
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019517312247246512,
+      "loss": 0.8895,
+      "step": 467
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019514618893187353,
+      "loss": 0.8736,
+      "step": 468
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001951191823257088,
+      "loss": 0.8701,
+      "step": 469
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019509210267471021,
+      "loss": 0.9189,
+      "step": 470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019506494999967298,
+      "loss": 0.8485,
+      "step": 471
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019503772432144845,
+      "loss": 0.8514,
+      "step": 472
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000195010425660944,
+      "loss": 0.8649,
+      "step": 473
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019498305403912314,
+      "loss": 0.8983,
+      "step": 474
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019495560947700532,
+      "loss": 0.2323,
+      "step": 475
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019492809199566597,
+      "loss": 0.8385,
+      "step": 476
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001949005016162367,
+      "loss": 0.881,
+      "step": 477
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019487283835990484,
+      "loss": 0.8547,
+      "step": 478
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019484510224791393,
+      "loss": 0.9215,
+      "step": 479
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019481729330156334,
+      "loss": 0.8363,
+      "step": 480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001947894115422083,
+      "loss": 0.8773,
+      "step": 481
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019476145699126013,
+      "loss": 0.9054,
+      "step": 482
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019473342967018593,
+      "loss": 0.8907,
+      "step": 483
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019470532960050866,
+      "loss": 0.8466,
+      "step": 484
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019467715680380728,
+      "loss": 0.8612,
+      "step": 485
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019464891130171647,
+      "loss": 0.8481,
+      "step": 486
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019462059311592676,
+      "loss": 0.8755,
+      "step": 487
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019459220226818457,
+      "loss": 0.9056,
+      "step": 488
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019456373878029202,
+      "loss": 0.8577,
+      "step": 489
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019453520267410712,
+      "loss": 0.8438,
+      "step": 490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019450659397154355,
+      "loss": 0.8659,
+      "step": 491
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001944779126945708,
+      "loss": 0.8224,
+      "step": 492
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019444915886521406,
+      "loss": 0.8509,
+      "step": 493
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019442033250555426,
+      "loss": 0.8701,
+      "step": 494
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019439143363772802,
+      "loss": 0.9098,
+      "step": 495
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019436246228392762,
+      "loss": 0.8929,
+      "step": 496
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019433341846640097,
+      "loss": 0.8712,
+      "step": 497
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019430430220745176,
+      "loss": 0.8805,
+      "step": 498
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001942751135294392,
+      "loss": 0.8781,
+      "step": 499
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019424585245477813,
+      "loss": 0.8862,
+      "step": 500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000194216519005939,
+      "loss": 0.8216,
+      "step": 501
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001941871132054478,
+      "loss": 0.8503,
+      "step": 502
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001941576350758862,
+      "loss": 0.8734,
+      "step": 503
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019412808463989121,
+      "loss": 0.8452,
+      "step": 504
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001940984619201556,
+      "loss": 0.8813,
+      "step": 505
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019406876693942745,
+      "loss": 0.8975,
+      "step": 506
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019403899972051045,
+      "loss": 0.8478,
+      "step": 507
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019400916028626376,
+      "loss": 0.8812,
+      "step": 508
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019397924865960185,
+      "loss": 0.8468,
+      "step": 509
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019394926486349486,
+      "loss": 0.878,
+      "step": 510
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001939192089209682,
+      "loss": 0.2398,
+      "step": 511
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019388908085510268,
+      "loss": 0.8839,
+      "step": 512
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001938588806890346,
+      "loss": 0.859,
+      "step": 513
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001938286084459555,
+      "loss": 0.9109,
+      "step": 514
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019379826414911238,
+      "loss": 0.8772,
+      "step": 515
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019376784782180746,
+      "loss": 0.9068,
+      "step": 516
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001937373594873984,
+      "loss": 0.8437,
+      "step": 517
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019370679916929807,
+      "loss": 0.8558,
+      "step": 518
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019367616689097467,
+      "loss": 0.8238,
+      "step": 519
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019364546267595153,
+      "loss": 0.8905,
+      "step": 520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019361468654780745,
+      "loss": 0.8599,
+      "step": 521
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019358383853017628,
+      "loss": 0.8446,
+      "step": 522
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019355291864674713,
+      "loss": 0.8789,
+      "step": 523
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019352192692126426,
+      "loss": 0.8793,
+      "step": 524
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019349086337752715,
+      "loss": 0.8948,
+      "step": 525
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019345972803939046,
+      "loss": 0.8692,
+      "step": 526
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001934285209307639,
+      "loss": 0.2478,
+      "step": 527
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019339724207561231,
+      "loss": 0.885,
+      "step": 528
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019336589149795567,
+      "loss": 0.8386,
+      "step": 529
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019333446922186904,
+      "loss": 0.8713,
+      "step": 530
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019330297527148246,
+      "loss": 0.877,
+      "step": 531
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019327140967098113,
+      "loss": 0.7951,
+      "step": 532
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019323977244460516,
+      "loss": 0.8592,
+      "step": 533
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019320806361664973,
+      "loss": 0.8572,
+      "step": 534
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000193176283211465,
+      "loss": 0.8329,
+      "step": 535
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019314443125345606,
+      "loss": 0.8572,
+      "step": 536
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000193112507767083,
+      "loss": 0.8994,
+      "step": 537
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019308051277686076,
+      "loss": 0.8709,
+      "step": 538
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001930484463073593,
+      "loss": 0.9168,
+      "step": 539
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019301630838320332,
+      "loss": 0.8668,
+      "step": 540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001929840990290726,
+      "loss": 0.8373,
+      "step": 541
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019295181826970158,
+      "loss": 0.8706,
+      "step": 542
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019291946612987962,
+      "loss": 0.8091,
+      "step": 543
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019288704263445089,
+      "loss": 0.2368,
+      "step": 544
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001928545478083144,
+      "loss": 0.8688,
+      "step": 545
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001928219816764238,
+      "loss": 0.871,
+      "step": 546
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019278934426378763,
+      "loss": 0.8914,
+      "step": 547
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019275663559546914,
+      "loss": 0.9077,
+      "step": 548
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019272385569658627,
+      "loss": 0.8967,
+      "step": 549
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019269100459231165,
+      "loss": 0.797,
+      "step": 550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019265808230787266,
+      "loss": 0.8474,
+      "step": 551
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019262508886855125,
+      "loss": 0.845,
+      "step": 552
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001925920242996841,
+      "loss": 0.8335,
+      "step": 553
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019255888862666244,
+      "loss": 0.8524,
+      "step": 554
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019252568187493215,
+      "loss": 0.8835,
+      "step": 555
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019249240406999366,
+      "loss": 0.8455,
+      "step": 556
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019245905523740202,
+      "loss": 0.8496,
+      "step": 557
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019242563540276675,
+      "loss": 0.8848,
+      "step": 558
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001923921445917519,
+      "loss": 0.8658,
+      "step": 559
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019235858283007612,
+      "loss": 0.8746,
+      "step": 560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019232495014351246,
+      "loss": 0.8584,
+      "step": 561
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019229124655788845,
+      "loss": 0.8902,
+      "step": 562
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019225747209908605,
+      "loss": 0.8973,
+      "step": 563
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001922236267930417,
+      "loss": 0.2532,
+      "step": 564
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019218971066574617,
+      "loss": 0.8559,
+      "step": 565
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019215572374324468,
+      "loss": 0.911,
+      "step": 566
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019212166605163682,
+      "loss": 0.835,
+      "step": 567
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019208753761707645,
+      "loss": 0.8991,
+      "step": 568
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019205333846577185,
+      "loss": 0.8642,
+      "step": 569
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019201906862398551,
+      "loss": 0.9046,
+      "step": 570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001919847281180343,
+      "loss": 0.824,
+      "step": 571
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001919503169742893,
+      "loss": 0.8472,
+      "step": 572
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019191583521917585,
+      "loss": 0.8858,
+      "step": 573
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019188128287917348,
+      "loss": 0.868,
+      "step": 574
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019184665998081598,
+      "loss": 0.8455,
+      "step": 575
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019181196655069127,
+      "loss": 0.8664,
+      "step": 576
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001917772026154415,
+      "loss": 0.8275,
+      "step": 577
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019174236820176292,
+      "loss": 0.8917,
+      "step": 578
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001917074633364059,
+      "loss": 0.8638,
+      "step": 579
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001916724880461749,
+      "loss": 0.84,
+      "step": 580
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019163744235792845,
+      "loss": 0.8754,
+      "step": 581
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019160232629857925,
+      "loss": 0.8919,
+      "step": 582
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019156713989509393,
+      "loss": 0.841,
+      "step": 583
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019153188317449314,
+      "loss": 0.8888,
+      "step": 584
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001914965561638516,
+      "loss": 0.871,
+      "step": 585
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019146115889029795,
+      "loss": 0.831,
+      "step": 586
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019142569138101478,
+      "loss": 0.8864,
+      "step": 587
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019139015366323868,
+      "loss": 0.8913,
+      "step": 588
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001913545457642601,
+      "loss": 0.847,
+      "step": 589
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001913188677114234,
+      "loss": 0.8548,
+      "step": 590
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019128311953212678,
+      "loss": 0.8515,
+      "step": 591
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019124730125382236,
+      "loss": 0.8543,
+      "step": 592
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019121141290401607,
+      "loss": 0.8856,
+      "step": 593
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001911754545102676,
+      "loss": 0.8546,
+      "step": 594
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019113942610019048,
+      "loss": 0.9226,
+      "step": 595
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000191103327701452,
+      "loss": 0.8801,
+      "step": 596
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019106715934177315,
+      "loss": 0.8895,
+      "step": 597
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019103092104892871,
+      "loss": 0.8189,
+      "step": 598
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019099461285074717,
+      "loss": 0.8778,
+      "step": 599
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019095823477511062,
+      "loss": 0.8773,
+      "step": 600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019092178684995489,
+      "loss": 0.8507,
+      "step": 601
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019088526910326938,
+      "loss": 0.8972,
+      "step": 602
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019084868156309717,
+      "loss": 0.8651,
+      "step": 603
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019081202425753495,
+      "loss": 0.879,
+      "step": 604
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001907752972147329,
+      "loss": 0.8634,
+      "step": 605
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019073850046289484,
+      "loss": 0.8848,
+      "step": 606
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019070163403027806,
+      "loss": 0.8288,
+      "step": 607
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019066469794519337,
+      "loss": 0.2526,
+      "step": 608
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001906276922360051,
+      "loss": 0.8803,
+      "step": 609
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019059061693113108,
+      "loss": 0.872,
+      "step": 610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019055347205904245,
+      "loss": 0.8749,
+      "step": 611
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001905162576482639,
+      "loss": 0.8786,
+      "step": 612
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019047897372737347,
+      "loss": 0.8479,
+      "step": 613
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019044162032500256,
+      "loss": 0.8248,
+      "step": 614
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000190404197469836,
+      "loss": 0.8662,
+      "step": 615
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019036670519061188,
+      "loss": 0.8601,
+      "step": 616
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019032914351612163,
+      "loss": 0.9047,
+      "step": 617
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019029151247520998,
+      "loss": 0.259,
+      "step": 618
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001902538120967749,
+      "loss": 0.8466,
+      "step": 619
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019021604240976764,
+      "loss": 0.2372,
+      "step": 620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001901782034431927,
+      "loss": 0.8314,
+      "step": 621
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019014029522610769,
+      "loss": 0.2338,
+      "step": 622
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001901023177876234,
+      "loss": 0.8286,
+      "step": 623
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019006427115690397,
+      "loss": 0.8537,
+      "step": 624
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001900261553631664,
+      "loss": 0.8293,
+      "step": 625
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018998797043568103,
+      "loss": 0.8588,
+      "step": 626
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018994971640377112,
+      "loss": 0.8603,
+      "step": 627
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018991139329681316,
+      "loss": 0.9031,
+      "step": 628
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001898730011442365,
+      "loss": 0.8835,
+      "step": 629
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018983453997552368,
+      "loss": 0.897,
+      "step": 630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018979600982021013,
+      "loss": 0.8234,
+      "step": 631
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018975741070788432,
+      "loss": 0.2862,
+      "step": 632
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018971874266818762,
+      "loss": 0.8899,
+      "step": 633
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001896800057308144,
+      "loss": 0.8944,
+      "step": 634
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001896411999255119,
+      "loss": 0.8625,
+      "step": 635
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018960232528208022,
+      "loss": 0.9003,
+      "step": 636
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018956338183037235,
+      "loss": 0.8162,
+      "step": 637
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018952436960029413,
+      "loss": 0.8598,
+      "step": 638
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001894852886218042,
+      "loss": 0.882,
+      "step": 639
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000189446138924914,
+      "loss": 0.8891,
+      "step": 640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018940692053968773,
+      "loss": 0.8443,
+      "step": 641
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018936763349624238,
+      "loss": 0.843,
+      "step": 642
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001893282778247476,
+      "loss": 0.8683,
+      "step": 643
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018928885355542577,
+      "loss": 0.8763,
+      "step": 644
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000189249360718552,
+      "loss": 0.8677,
+      "step": 645
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000189209799344454,
+      "loss": 0.8541,
+      "step": 646
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018917016946351205,
+      "loss": 0.9029,
+      "step": 647
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018913047110615916,
+      "loss": 0.8563,
+      "step": 648
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018909070430288088,
+      "loss": 0.9013,
+      "step": 649
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001890508690842153,
+      "loss": 0.845,
+      "step": 650
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018901096548075305,
+      "loss": 0.8852,
+      "step": 651
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018897099352313732,
+      "loss": 0.8414,
+      "step": 652
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018893095324206373,
+      "loss": 0.8674,
+      "step": 653
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018889084466828043,
+      "loss": 0.877,
+      "step": 654
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001888506678325879,
+      "loss": 0.9062,
+      "step": 655
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018881042276583924,
+      "loss": 0.8285,
+      "step": 656
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018877010949893975,
+      "loss": 0.8635,
+      "step": 657
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018872972806284718,
+      "loss": 0.8668,
+      "step": 658
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018868927848857166,
+      "loss": 0.8465,
+      "step": 659
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018864876080717561,
+      "loss": 0.9042,
+      "step": 660
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018860817504977372,
+      "loss": 0.8758,
+      "step": 661
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018856752124753304,
+      "loss": 0.8918,
+      "step": 662
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001885267994316728,
+      "loss": 0.8938,
+      "step": 663
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001884860096334645,
+      "loss": 0.8817,
+      "step": 664
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001884451518842318,
+      "loss": 0.2596,
+      "step": 665
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018840422621535066,
+      "loss": 0.8806,
+      "step": 666
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000188363232658249,
+      "loss": 0.2554,
+      "step": 667
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018832217124440702,
+      "loss": 0.8731,
+      "step": 668
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001882810420053571,
+      "loss": 0.8445,
+      "step": 669
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018823984497268344,
+      "loss": 0.2439,
+      "step": 670
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018819858017802253,
+      "loss": 0.8782,
+      "step": 671
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018815724765306287,
+      "loss": 0.8695,
+      "step": 672
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018811584742954484,
+      "loss": 0.829,
+      "step": 673
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018807437953926096,
+      "loss": 0.8599,
+      "step": 674
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018803284401405564,
+      "loss": 0.8689,
+      "step": 675
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001879912408858252,
+      "loss": 0.8835,
+      "step": 676
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000187949570186518,
+      "loss": 0.8599,
+      "step": 677
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018790783194813412,
+      "loss": 0.8416,
+      "step": 678
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001878660262027256,
+      "loss": 0.8754,
+      "step": 679
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018782415298239635,
+      "loss": 0.8807,
+      "step": 680
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018778221231930203,
+      "loss": 0.8418,
+      "step": 681
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001877402042456501,
+      "loss": 0.8771,
+      "step": 682
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001876981287936998,
+      "loss": 0.8912,
+      "step": 683
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018765598599576215,
+      "loss": 0.8572,
+      "step": 684
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018761377588419985,
+      "loss": 0.8673,
+      "step": 685
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018757149849142724,
+      "loss": 0.8973,
+      "step": 686
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018752915384991043,
+      "loss": 0.9198,
+      "step": 687
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018748674199216712,
+      "loss": 0.2833,
+      "step": 688
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001874442629507666,
+      "loss": 0.8559,
+      "step": 689
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018740171675832982,
+      "loss": 0.8293,
+      "step": 690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018735910344752922,
+      "loss": 0.8952,
+      "step": 691
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018731642305108885,
+      "loss": 0.8458,
+      "step": 692
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018727367560178424,
+      "loss": 0.8496,
+      "step": 693
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018723086113244243,
+      "loss": 0.8311,
+      "step": 694
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018718797967594191,
+      "loss": 0.9,
+      "step": 695
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001871450312652126,
+      "loss": 0.8644,
+      "step": 696
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001871020159332359,
+      "loss": 0.8357,
+      "step": 697
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018705893371304453,
+      "loss": 0.8558,
+      "step": 698
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001870157846377226,
+      "loss": 0.8628,
+      "step": 699
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018697256874040553,
+      "loss": 0.8228,
+      "step": 700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018692928605428016,
+      "loss": 0.8644,
+      "step": 701
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001868859366125845,
+      "loss": 0.8812,
+      "step": 702
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018684252044860785,
+      "loss": 0.8529,
+      "step": 703
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001867990375956908,
+      "loss": 0.9216,
+      "step": 704
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001867554880872251,
+      "loss": 0.8869,
+      "step": 705
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001867118719566537,
+      "loss": 0.8217,
+      "step": 706
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018666818923747077,
+      "loss": 0.8289,
+      "step": 707
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001866244399632215,
+      "loss": 0.9035,
+      "step": 708
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001865806241675023,
+      "loss": 0.828,
+      "step": 709
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018653674188396057,
+      "loss": 0.8598,
+      "step": 710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018649279314629483,
+      "loss": 0.8262,
+      "step": 711
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018644877798825463,
+      "loss": 0.825,
+      "step": 712
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018640469644364048,
+      "loss": 0.8719,
+      "step": 713
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018636054854630395,
+      "loss": 0.8893,
+      "step": 714
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001863163343301475,
+      "loss": 0.8665,
+      "step": 715
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001862720538291245,
+      "loss": 0.8838,
+      "step": 716
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018622770707723928,
+      "loss": 0.866,
+      "step": 717
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018618329410854704,
+      "loss": 0.878,
+      "step": 718
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018613881495715376,
+      "loss": 0.8695,
+      "step": 719
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018609426965721638,
+      "loss": 0.8711,
+      "step": 720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018604965824294253,
+      "loss": 0.8672,
+      "step": 721
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018600498074859055,
+      "loss": 0.8242,
+      "step": 722
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018596023720846968,
+      "loss": 0.8635,
+      "step": 723
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018591542765693982,
+      "loss": 0.9149,
+      "step": 724
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018587055212841152,
+      "loss": 0.8636,
+      "step": 725
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018582561065734604,
+      "loss": 0.8493,
+      "step": 726
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018578060327825523,
+      "loss": 0.8943,
+      "step": 727
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018573553002570166,
+      "loss": 0.8924,
+      "step": 728
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018569039093429835,
+      "loss": 0.8619,
+      "step": 729
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018564518603870895,
+      "loss": 0.8597,
+      "step": 730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018559991537364767,
+      "loss": 0.8652,
+      "step": 731
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001855545789738792,
+      "loss": 0.8828,
+      "step": 732
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018550917687421866,
+      "loss": 0.8955,
+      "step": 733
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018546370910953167,
+      "loss": 0.8408,
+      "step": 734
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018541817571473429,
+      "loss": 0.8577,
+      "step": 735
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018537257672479292,
+      "loss": 0.8175,
+      "step": 736
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018532691217472446,
+      "loss": 0.8425,
+      "step": 737
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018528118209959599,
+      "loss": 0.9016,
+      "step": 738
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018523538653452505,
+      "loss": 0.8799,
+      "step": 739
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018518952551467934,
+      "loss": 0.2767,
+      "step": 740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001851435990752769,
+      "loss": 0.8105,
+      "step": 741
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001850976072515861,
+      "loss": 0.8841,
+      "step": 742
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001850515500789253,
+      "loss": 0.8536,
+      "step": 743
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018500542759266323,
+      "loss": 0.9186,
+      "step": 744
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018495923982821873,
+      "loss": 0.2406,
+      "step": 745
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018491298682106065,
+      "loss": 0.8463,
+      "step": 746
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018486666860670817,
+      "loss": 0.8728,
+      "step": 747
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018482028522073032,
+      "loss": 0.8431,
+      "step": 748
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018477383669874634,
+      "loss": 0.8551,
+      "step": 749
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001847273230764254,
+      "loss": 0.2244,
+      "step": 750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018468074438948663,
+      "loss": 0.8843,
+      "step": 751
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001846341006736993,
+      "loss": 0.8669,
+      "step": 752
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001845873919648824,
+      "loss": 0.8399,
+      "step": 753
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018454061829890496,
+      "loss": 0.8528,
+      "step": 754
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018449377971168588,
+      "loss": 0.8438,
+      "step": 755
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018444687623919386,
+      "loss": 0.8358,
+      "step": 756
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001843999079174475,
+      "loss": 0.852,
+      "step": 757
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018435287478251516,
+      "loss": 0.8388,
+      "step": 758
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018430577687051494,
+      "loss": 0.2407,
+      "step": 759
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018425861421761475,
+      "loss": 0.8551,
+      "step": 760
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001842113868600322,
+      "loss": 0.8664,
+      "step": 761
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018416409483403454,
+      "loss": 0.7996,
+      "step": 762
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018411673817593873,
+      "loss": 0.8366,
+      "step": 763
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018406931692211136,
+      "loss": 0.8899,
+      "step": 764
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018402183110896857,
+      "loss": 0.8825,
+      "step": 765
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001839742807729762,
+      "loss": 0.879,
+      "step": 766
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018392666595064947,
+      "loss": 0.849,
+      "step": 767
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001838789866785533,
+      "loss": 0.813,
+      "step": 768
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018383124299330193,
+      "loss": 0.8637,
+      "step": 769
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018378343493155922,
+      "loss": 0.873,
+      "step": 770
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001837355625300383,
+      "loss": 0.8831,
+      "step": 771
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018368762582550194,
+      "loss": 0.8533,
+      "step": 772
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018363962485476204,
+      "loss": 0.9366,
+      "step": 773
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018359155965468,
+      "loss": 0.8851,
+      "step": 774
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018354343026216654,
+      "loss": 0.8252,
+      "step": 775
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018349523671418159,
+      "loss": 0.8147,
+      "step": 776
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018344697904773444,
+      "loss": 0.817,
+      "step": 777
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018339865729988356,
+      "loss": 0.8623,
+      "step": 778
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018335027150773666,
+      "loss": 0.8656,
+      "step": 779
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001833018217084506,
+      "loss": 0.8986,
+      "step": 780
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018325330793923144,
+      "loss": 0.8609,
+      "step": 781
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018320473023733435,
+      "loss": 0.8552,
+      "step": 782
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018315608864006354,
+      "loss": 0.8079,
+      "step": 783
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018310738318477236,
+      "loss": 0.8756,
+      "step": 784
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001830586139088632,
+      "loss": 0.8587,
+      "step": 785
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018300978084978735,
+      "loss": 0.8889,
+      "step": 786
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018296088404504523,
+      "loss": 0.8218,
+      "step": 787
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018291192353218612,
+      "loss": 0.8924,
+      "step": 788
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018286289934880826,
+      "loss": 0.8613,
+      "step": 789
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018281381153255872,
+      "loss": 0.8628,
+      "step": 790
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018276466012113356,
+      "loss": 0.8473,
+      "step": 791
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018271544515227755,
+      "loss": 0.8174,
+      "step": 792
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001826661666637843,
+      "loss": 0.8389,
+      "step": 793
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018261682469349627,
+      "loss": 0.9194,
+      "step": 794
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018256741927930462,
+      "loss": 0.8248,
+      "step": 795
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018251795045914922,
+      "loss": 0.8089,
+      "step": 796
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018246841827101863,
+      "loss": 0.9071,
+      "step": 797
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001824188227529501,
+      "loss": 0.8482,
+      "step": 798
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018236916394302944,
+      "loss": 0.8199,
+      "step": 799
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018231944187939116,
+      "loss": 0.7998,
+      "step": 800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018226965660021836,
+      "loss": 0.8255,
+      "step": 801
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001822198081437425,
+      "loss": 0.8805,
+      "step": 802
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001821698965482438,
+      "loss": 0.8539,
+      "step": 803
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001821199218520508,
+      "loss": 0.8942,
+      "step": 804
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018206988409354054,
+      "loss": 0.8819,
+      "step": 805
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018201978331113853,
+      "loss": 0.7955,
+      "step": 806
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018196961954331862,
+      "loss": 0.8231,
+      "step": 807
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018191939282860307,
+      "loss": 0.8494,
+      "step": 808
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001818691032055624,
+      "loss": 0.8415,
+      "step": 809
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018181875071281555,
+      "loss": 0.8339,
+      "step": 810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018176833538902968,
+      "loss": 0.8588,
+      "step": 811
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018171785727292026,
+      "loss": 0.872,
+      "step": 812
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018166731640325084,
+      "loss": 0.8375,
+      "step": 813
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001816167128188333,
+      "loss": 0.8417,
+      "step": 814
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001815660465585276,
+      "loss": 0.8308,
+      "step": 815
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018151531766124186,
+      "loss": 0.8644,
+      "step": 816
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018146452616593235,
+      "loss": 0.8842,
+      "step": 817
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001814136721116033,
+      "loss": 0.8249,
+      "step": 818
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001813627555373071,
+      "loss": 0.8776,
+      "step": 819
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018131177648214407,
+      "loss": 0.8466,
+      "step": 820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018126073498526255,
+      "loss": 0.8688,
+      "step": 821
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018120963108585872,
+      "loss": 0.8666,
+      "step": 822
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001811584648231769,
+      "loss": 0.8583,
+      "step": 823
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018110723623650915,
+      "loss": 0.8651,
+      "step": 824
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001810559453651954,
+      "loss": 0.8626,
+      "step": 825
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018100459224862337,
+      "loss": 0.2339,
+      "step": 826
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018095317692622871,
+      "loss": 0.8197,
+      "step": 827
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018090169943749476,
+      "loss": 0.2253,
+      "step": 828
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018085015982195258,
+      "loss": 0.8423,
+      "step": 829
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018079855811918097,
+      "loss": 0.8115,
+      "step": 830
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018074689436880644,
+      "loss": 0.8692,
+      "step": 831
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018069516861050309,
+      "loss": 0.8547,
+      "step": 832
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018064338088399264,
+      "loss": 0.8599,
+      "step": 833
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018059153122904442,
+      "loss": 0.2578,
+      "step": 834
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018053961968547534,
+      "loss": 0.8924,
+      "step": 835
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018048764629314978,
+      "loss": 0.8318,
+      "step": 836
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018043561109197967,
+      "loss": 0.8625,
+      "step": 837
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001803835141219244,
+      "loss": 0.8379,
+      "step": 838
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001803313554229907,
+      "loss": 0.8928,
+      "step": 839
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018027913503523287,
+      "loss": 0.8352,
+      "step": 840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018022685299875243,
+      "loss": 0.8542,
+      "step": 841
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001801745093536983,
+      "loss": 0.8586,
+      "step": 842
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018012210414026676,
+      "loss": 0.8166,
+      "step": 843
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001800696373987013,
+      "loss": 0.8949,
+      "step": 844
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018001710916929262,
+      "loss": 0.8451,
+      "step": 845
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001799645194923788,
+      "loss": 0.8676,
+      "step": 846
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017991186840834496,
+      "loss": 0.8387,
+      "step": 847
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017985915595762343,
+      "loss": 0.8397,
+      "step": 848
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017980638218069366,
+      "loss": 0.85,
+      "step": 849
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001797535471180822,
+      "loss": 0.8641,
+      "step": 850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017970065081036264,
+      "loss": 0.8528,
+      "step": 851
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017964769329815568,
+      "loss": 0.8387,
+      "step": 852
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017959467462212887,
+      "loss": 0.8559,
+      "step": 853
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017954159482299687,
+      "loss": 0.8265,
+      "step": 854
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017948845394152123,
+      "loss": 0.8755,
+      "step": 855
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001794352520185104,
+      "loss": 0.8353,
+      "step": 856
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001793819890948197,
+      "loss": 0.8556,
+      "step": 857
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001793286652113513,
+      "loss": 0.8396,
+      "step": 858
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017927528040905425,
+      "loss": 0.812,
+      "step": 859
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001792218347289242,
+      "loss": 0.8607,
+      "step": 860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017916832821200375,
+      "loss": 0.8653,
+      "step": 861
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017911476089938218,
+      "loss": 0.8459,
+      "step": 862
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001790611328321953,
+      "loss": 0.8434,
+      "step": 863
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017900744405162578,
+      "loss": 0.8483,
+      "step": 864
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017895369459890276,
+      "loss": 0.8204,
+      "step": 865
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001788998845153021,
+      "loss": 0.878,
+      "step": 866
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017884601384214607,
+      "loss": 0.9063,
+      "step": 867
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017879208262080364,
+      "loss": 0.8802,
+      "step": 868
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017873809089269017,
+      "loss": 0.8863,
+      "step": 869
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017868403869926748,
+      "loss": 0.8653,
+      "step": 870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017862992608204384,
+      "loss": 0.8704,
+      "step": 871
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017857575308257394,
+      "loss": 0.8116,
+      "step": 872
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017852151974245882,
+      "loss": 0.8939,
+      "step": 873
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017846722610334592,
+      "loss": 0.8562,
+      "step": 874
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017841287220692889,
+      "loss": 0.8265,
+      "step": 875
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017835845809494768,
+      "loss": 0.819,
+      "step": 876
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017830398380918854,
+      "loss": 0.863,
+      "step": 877
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017824944939148383,
+      "loss": 0.8207,
+      "step": 878
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001781948548837122,
+      "loss": 0.838,
+      "step": 879
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017814020032779832,
+      "loss": 0.8988,
+      "step": 880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017808548576571314,
+      "loss": 0.8577,
+      "step": 881
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001780307112394735,
+      "loss": 0.8453,
+      "step": 882
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001779758767911424,
+      "loss": 0.8659,
+      "step": 883
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017792098246282884,
+      "loss": 0.8162,
+      "step": 884
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017786602829668782,
+      "loss": 0.8539,
+      "step": 885
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017781101433492024,
+      "loss": 0.8244,
+      "step": 886
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000177755940619773,
+      "loss": 0.8254,
+      "step": 887
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017770080719353874,
+      "loss": 0.9128,
+      "step": 888
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017764561409855617,
+      "loss": 0.8534,
+      "step": 889
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017759036137720958,
+      "loss": 0.8613,
+      "step": 890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017753504907192923,
+      "loss": 0.8512,
+      "step": 891
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001774796772251911,
+      "loss": 0.8189,
+      "step": 892
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017742424587951683,
+      "loss": 0.8228,
+      "step": 893
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017736875507747379,
+      "loss": 0.834,
+      "step": 894
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017731320486167503,
+      "loss": 0.8463,
+      "step": 895
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001772575952747792,
+      "loss": 0.8434,
+      "step": 896
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001772019263594905,
+      "loss": 0.8781,
+      "step": 897
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017714619815855883,
+      "loss": 0.8714,
+      "step": 898
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017709041071477946,
+      "loss": 0.8248,
+      "step": 899
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001770345640709932,
+      "loss": 0.8883,
+      "step": 900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001769786582700864,
+      "loss": 0.8742,
+      "step": 901
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001769226933549907,
+      "loss": 0.8713,
+      "step": 902
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001768666693686833,
+      "loss": 0.874,
+      "step": 903
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017681058635418662,
+      "loss": 0.8397,
+      "step": 904
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001767544443545685,
+      "loss": 0.8849,
+      "step": 905
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017669824341294202,
+      "loss": 0.8143,
+      "step": 906
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017664198357246553,
+      "loss": 0.8145,
+      "step": 907
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017658566487634267,
+      "loss": 0.8101,
+      "step": 908
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001765292873678222,
+      "loss": 0.8858,
+      "step": 909
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001764728510901981,
+      "loss": 0.2427,
+      "step": 910
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001764163560868094,
+      "loss": 0.8849,
+      "step": 911
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001763598024010404,
+      "loss": 0.2887,
+      "step": 912
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001763031900763202,
+      "loss": 0.8527,
+      "step": 913
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001762465191561232,
+      "loss": 0.8608,
+      "step": 914
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017618978968396865,
+      "loss": 0.8603,
+      "step": 915
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017613300170342072,
+      "loss": 0.8716,
+      "step": 916
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017607615525808864,
+      "loss": 0.8169,
+      "step": 917
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017601925039162648,
+      "loss": 0.8179,
+      "step": 918
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017596228714773316,
+      "loss": 0.8815,
+      "step": 919
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017590526557015242,
+      "loss": 0.8494,
+      "step": 920
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017584818570267284,
+      "loss": 0.8518,
+      "step": 921
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017579104758912777,
+      "loss": 0.8489,
+      "step": 922
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001757338512733952,
+      "loss": 0.8482,
+      "step": 923
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017567659679939787,
+      "loss": 0.8595,
+      "step": 924
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017561928421110323,
+      "loss": 0.8703,
+      "step": 925
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017556191355252328,
+      "loss": 0.8499,
+      "step": 926
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017550448486771466,
+      "loss": 0.8448,
+      "step": 927
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017544699820077855,
+      "loss": 0.8843,
+      "step": 928
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017538945359586066,
+      "loss": 0.821,
+      "step": 929
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001753318510971512,
+      "loss": 0.8919,
+      "step": 930
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017527419074888482,
+      "loss": 0.8512,
+      "step": 931
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017521647259534064,
+      "loss": 0.8383,
+      "step": 932
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017515869668084208,
+      "loss": 0.8525,
+      "step": 933
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017510086304975704,
+      "loss": 0.8409,
+      "step": 934
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017504297174649765,
+      "loss": 0.7845,
+      "step": 935
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001749850228155203,
+      "loss": 0.8213,
+      "step": 936
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017492701630132578,
+      "loss": 0.8273,
+      "step": 937
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017486895224845897,
+      "loss": 0.8444,
+      "step": 938
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017481083070150895,
+      "loss": 0.8733,
+      "step": 939
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000174752651705109,
+      "loss": 0.8836,
+      "step": 940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001746944153039365,
+      "loss": 0.8602,
+      "step": 941
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001746361215427129,
+      "loss": 0.8541,
+      "step": 942
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017457777046620367,
+      "loss": 0.7916,
+      "step": 943
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017451936211921836,
+      "loss": 0.8834,
+      "step": 944
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017446089654661048,
+      "loss": 0.8686,
+      "step": 945
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017440237379327743,
+      "loss": 0.8639,
+      "step": 946
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001743437939041606,
+      "loss": 0.8623,
+      "step": 947
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001742851569242452,
+      "loss": 0.8261,
+      "step": 948
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017422646289856034,
+      "loss": 0.8013,
+      "step": 949
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017416771187217883,
+      "loss": 0.8854,
+      "step": 950
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017410890389021736,
+      "loss": 0.8795,
+      "step": 951
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017405003899783635,
+      "loss": 0.8308,
+      "step": 952
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001739911172402398,
+      "loss": 0.8473,
+      "step": 953
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001739321386626756,
+      "loss": 0.8539,
+      "step": 954
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017387310331043503,
+      "loss": 0.8005,
+      "step": 955
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017381401122885313,
+      "loss": 0.8716,
+      "step": 956
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017375486246330848,
+      "loss": 0.8481,
+      "step": 957
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017369565705922313,
+      "loss": 0.8615,
+      "step": 958
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001736363950620627,
+      "loss": 0.8736,
+      "step": 959
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017357707651733618,
+      "loss": 0.8126,
+      "step": 960
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017351770147059603,
+      "loss": 0.8501,
+      "step": 961
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017345826996743817,
+      "loss": 0.82,
+      "step": 962
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017339878205350177,
+      "loss": 0.8982,
+      "step": 963
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017333923777446936,
+      "loss": 0.8118,
+      "step": 964
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017327963717606677,
+      "loss": 0.8402,
+      "step": 965
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000173219980304063,
+      "loss": 0.8691,
+      "step": 966
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017316026720427043,
+      "loss": 0.8348,
+      "step": 967
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017310049792254443,
+      "loss": 0.8506,
+      "step": 968
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017304067250478364,
+      "loss": 0.8505,
+      "step": 969
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017298079099692978,
+      "loss": 0.857,
+      "step": 970
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017292085344496757,
+      "loss": 0.8187,
+      "step": 971
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001728608598949249,
+      "loss": 0.8259,
+      "step": 972
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017280081039287254,
+      "loss": 0.843,
+      "step": 973
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017274070498492434,
+      "loss": 0.8797,
+      "step": 974
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017268054371723693,
+      "loss": 0.8475,
+      "step": 975
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017262032663601003,
+      "loss": 0.8101,
+      "step": 976
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000172560053787486,
+      "loss": 0.8649,
+      "step": 977
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017249972521795023,
+      "loss": 0.8059,
+      "step": 978
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001724393409737308,
+      "loss": 0.2303,
+      "step": 979
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001723789011011986,
+      "loss": 0.8579,
+      "step": 980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001723184056467671,
+      "loss": 0.8471,
+      "step": 981
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017225785465689258,
+      "loss": 0.8178,
+      "step": 982
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017219724817807398,
+      "loss": 0.8878,
+      "step": 983
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017213658625685275,
+      "loss": 0.8617,
+      "step": 984
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017207586893981306,
+      "loss": 0.8463,
+      "step": 985
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017201509627358142,
+      "loss": 0.8415,
+      "step": 986
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017195426830482705,
+      "loss": 0.877,
+      "step": 987
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001718933850802615,
+      "loss": 0.8447,
+      "step": 988
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017183244664663877,
+      "loss": 0.827,
+      "step": 989
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001717714530507554,
+      "loss": 0.8934,
+      "step": 990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017171040433945005,
+      "loss": 0.8336,
+      "step": 991
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001716493005596039,
+      "loss": 0.2317,
+      "step": 992
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017158814175814033,
+      "loss": 0.8754,
+      "step": 993
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017152692798202501,
+      "loss": 0.8399,
+      "step": 994
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017146565927826578,
+      "loss": 0.8429,
+      "step": 995
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017140433569391275,
+      "loss": 0.2361,
+      "step": 996
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000171342957276058,
+      "loss": 0.9152,
+      "step": 997
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017128152407183597,
+      "loss": 0.8004,
+      "step": 998
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017122003612842296,
+      "loss": 0.8548,
+      "step": 999
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017115849349303735,
+      "loss": 0.8684,
+      "step": 1000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001710968962129396,
+      "loss": 0.8952,
+      "step": 1001
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001710352443354321,
+      "loss": 0.8664,
+      "step": 1002
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017097353790785912,
+      "loss": 0.8557,
+      "step": 1003
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017091177697760684,
+      "loss": 0.8291,
+      "step": 1004
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017084996159210332,
+      "loss": 0.253,
+      "step": 1005
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017078809179881844,
+      "loss": 0.8666,
+      "step": 1006
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001707261676452638,
+      "loss": 0.8636,
+      "step": 1007
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017066418917899284,
+      "loss": 0.818,
+      "step": 1008
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017060215644760065,
+      "loss": 0.2579,
+      "step": 1009
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017054006949872397,
+      "loss": 0.2408,
+      "step": 1010
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001704779283800412,
+      "loss": 0.8113,
+      "step": 1011
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001704157331392724,
+      "loss": 0.8428,
+      "step": 1012
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017035348382417906,
+      "loss": 0.8298,
+      "step": 1013
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017029118048256435,
+      "loss": 0.8476,
+      "step": 1014
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017022882316227274,
+      "loss": 0.8332,
+      "step": 1015
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001701664119111904,
+      "loss": 0.2134,
+      "step": 1016
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001701039467772447,
+      "loss": 0.8443,
+      "step": 1017
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017004142780840443,
+      "loss": 0.8305,
+      "step": 1018
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016997885505267977,
+      "loss": 0.8439,
+      "step": 1019
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001699162285581223,
+      "loss": 0.8591,
+      "step": 1020
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001698535483728246,
+      "loss": 0.8821,
+      "step": 1021
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016979081454492073,
+      "loss": 0.8221,
+      "step": 1022
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016972802712258585,
+      "loss": 0.8235,
+      "step": 1023
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016966518615403625,
+      "loss": 0.8547,
+      "step": 1024
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001696022916875294,
+      "loss": 0.8336,
+      "step": 1025
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016953934377136377,
+      "loss": 0.8243,
+      "step": 1026
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016947634245387895,
+      "loss": 0.8197,
+      "step": 1027
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016941328778345555,
+      "loss": 0.8493,
+      "step": 1028
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016935017980851508,
+      "loss": 0.8119,
+      "step": 1029
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016928701857752006,
+      "loss": 0.8419,
+      "step": 1030
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001692238041389738,
+      "loss": 0.8478,
+      "step": 1031
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001691605365414206,
+      "loss": 0.8051,
+      "step": 1032
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001690972158334455,
+      "loss": 0.8597,
+      "step": 1033
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016903384206367434,
+      "loss": 0.8543,
+      "step": 1034
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016897041528077372,
+      "loss": 0.8275,
+      "step": 1035
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001689069355334509,
+      "loss": 0.8416,
+      "step": 1036
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016884340287045394,
+      "loss": 0.8433,
+      "step": 1037
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016877981734057137,
+      "loss": 0.8159,
+      "step": 1038
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016871617899263242,
+      "loss": 0.8783,
+      "step": 1039
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001686524878755069,
+      "loss": 0.8349,
+      "step": 1040
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016858874403810509,
+      "loss": 0.8353,
+      "step": 1041
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016852494752937771,
+      "loss": 0.8343,
+      "step": 1042
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016846109839831606,
+      "loss": 0.8452,
+      "step": 1043
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016839719669395174,
+      "loss": 0.8305,
+      "step": 1044
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016833324246535677,
+      "loss": 0.804,
+      "step": 1045
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001682692357616435,
+      "loss": 0.8337,
+      "step": 1046
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016820517663196455,
+      "loss": 0.8483,
+      "step": 1047
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016814106512551283,
+      "loss": 0.8861,
+      "step": 1048
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016807690129152145,
+      "loss": 0.8434,
+      "step": 1049
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016801268517926377,
+      "loss": 0.8698,
+      "step": 1050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016794841683805318,
+      "loss": 0.8626,
+      "step": 1051
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016788409631724328,
+      "loss": 0.833,
+      "step": 1052
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016781972366622765,
+      "loss": 0.265,
+      "step": 1053
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016775529893444,
+      "loss": 0.8551,
+      "step": 1054
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016769082217135396,
+      "loss": 0.853,
+      "step": 1055
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016762629342648318,
+      "loss": 0.8509,
+      "step": 1056
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001675617127493812,
+      "loss": 0.2316,
+      "step": 1057
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016749708018964137,
+      "loss": 0.8347,
+      "step": 1058
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016743239579689699,
+      "loss": 0.8675,
+      "step": 1059
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016736765962082107,
+      "loss": 0.8696,
+      "step": 1060
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016730287171112652,
+      "loss": 0.2515,
+      "step": 1061
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001672380321175658,
+      "loss": 0.8531,
+      "step": 1062
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001671731408899312,
+      "loss": 0.8264,
+      "step": 1063
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016710819807805462,
+      "loss": 0.8354,
+      "step": 1064
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016704320373180754,
+      "loss": 0.8045,
+      "step": 1065
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001669781579011011,
+      "loss": 0.833,
+      "step": 1066
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016691306063588583,
+      "loss": 0.8616,
+      "step": 1067
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001668479119861519,
+      "loss": 0.8104,
+      "step": 1068
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016678271200192888,
+      "loss": 0.8435,
+      "step": 1069
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016671746073328576,
+      "loss": 0.842,
+      "step": 1070
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001666521582303309,
+      "loss": 0.2703,
+      "step": 1071
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016658680454321206,
+      "loss": 0.8716,
+      "step": 1072
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016652139972211626,
+      "loss": 0.8738,
+      "step": 1073
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016645594381726979,
+      "loss": 0.8574,
+      "step": 1074
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016639043687893815,
+      "loss": 0.8797,
+      "step": 1075
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016632487895742613,
+      "loss": 0.836,
+      "step": 1076
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001662592701030775,
+      "loss": 0.8623,
+      "step": 1077
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001661936103662753,
+      "loss": 0.2381,
+      "step": 1078
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001661278997974416,
+      "loss": 0.881,
+      "step": 1079
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016606213844703744,
+      "loss": 0.8564,
+      "step": 1080
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016599632636556293,
+      "loss": 0.8457,
+      "step": 1081
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016593046360355713,
+      "loss": 0.8425,
+      "step": 1082
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016586455021159799,
+      "loss": 0.8498,
+      "step": 1083
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016579858624030233,
+      "loss": 0.8501,
+      "step": 1084
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016573257174032582,
+      "loss": 0.8483,
+      "step": 1085
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016566650676236305,
+      "loss": 0.8292,
+      "step": 1086
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016560039135714717,
+      "loss": 0.8517,
+      "step": 1087
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016553422557545022,
+      "loss": 0.8544,
+      "step": 1088
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001654680094680828,
+      "loss": 0.8272,
+      "step": 1089
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016540174308589425,
+      "loss": 0.8731,
+      "step": 1090
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001653354264797725,
+      "loss": 0.8668,
+      "step": 1091
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016526905970064396,
+      "loss": 0.8551,
+      "step": 1092
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016520264279947376,
+      "loss": 0.8457,
+      "step": 1093
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016513617582726532,
+      "loss": 0.8552,
+      "step": 1094
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016506965883506065,
+      "loss": 0.7953,
+      "step": 1095
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016500309187394003,
+      "loss": 0.7985,
+      "step": 1096
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016493647499502226,
+      "loss": 0.8342,
+      "step": 1097
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016486980824946445,
+      "loss": 0.8309,
+      "step": 1098
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016480309168846188,
+      "loss": 0.8555,
+      "step": 1099
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016473632536324821,
+      "loss": 0.924,
+      "step": 1100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001646695093250953,
+      "loss": 0.8306,
+      "step": 1101
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016460264362531311,
+      "loss": 0.812,
+      "step": 1102
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016453572831524985,
+      "loss": 0.8906,
+      "step": 1103
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016446876344629172,
+      "loss": 0.8931,
+      "step": 1104
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016440174906986302,
+      "loss": 0.8873,
+      "step": 1105
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001643346852374261,
+      "loss": 0.8573,
+      "step": 1106
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016426757200048125,
+      "loss": 0.8385,
+      "step": 1107
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001642004094105667,
+      "loss": 0.8462,
+      "step": 1108
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001641331975192586,
+      "loss": 0.8533,
+      "step": 1109
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016406593637817101,
+      "loss": 0.8641,
+      "step": 1110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016399862603895562,
+      "loss": 0.8733,
+      "step": 1111
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001639312665533021,
+      "loss": 0.8579,
+      "step": 1112
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016386385797293783,
+      "loss": 0.8188,
+      "step": 1113
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001637964003496278,
+      "loss": 0.8821,
+      "step": 1114
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016372889373517474,
+      "loss": 0.8607,
+      "step": 1115
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016366133818141893,
+      "loss": 0.8593,
+      "step": 1116
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016359373374023832,
+      "loss": 0.2484,
+      "step": 1117
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001635260804635484,
+      "loss": 0.8543,
+      "step": 1118
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000163458378403302,
+      "loss": 0.8654,
+      "step": 1119
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016339062761148968,
+      "loss": 0.8382,
+      "step": 1120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016332282814013918,
+      "loss": 0.8475,
+      "step": 1121
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016325498004131575,
+      "loss": 0.8723,
+      "step": 1122
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016318708336712193,
+      "loss": 0.8571,
+      "step": 1123
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016311913816969762,
+      "loss": 0.8678,
+      "step": 1124
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016305114450121993,
+      "loss": 0.8137,
+      "step": 1125
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016298310241390324,
+      "loss": 0.851,
+      "step": 1126
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016291501195999904,
+      "loss": 0.8388,
+      "step": 1127
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000162846873191796,
+      "loss": 0.8843,
+      "step": 1128
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001627786861616199,
+      "loss": 0.7863,
+      "step": 1129
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001627104509218336,
+      "loss": 0.2383,
+      "step": 1130
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016264216752483697,
+      "loss": 0.8051,
+      "step": 1131
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001625738360230668,
+      "loss": 0.8484,
+      "step": 1132
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016250545646899695,
+      "loss": 0.8101,
+      "step": 1133
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000162437028915138,
+      "loss": 0.8125,
+      "step": 1134
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016236855341403755,
+      "loss": 0.8629,
+      "step": 1135
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016230003001827998,
+      "loss": 0.8411,
+      "step": 1136
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016223145878048643,
+      "loss": 0.8377,
+      "step": 1137
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016216283975331472,
+      "loss": 0.7942,
+      "step": 1138
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016209417298945953,
+      "loss": 0.8681,
+      "step": 1139
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000162025458541652,
+      "loss": 0.834,
+      "step": 1140
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016195669646266002,
+      "loss": 0.2366,
+      "step": 1141
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016188788680528807,
+      "loss": 0.8282,
+      "step": 1142
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016181902962237706,
+      "loss": 0.8257,
+      "step": 1143
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001617501249668045,
+      "loss": 0.8782,
+      "step": 1144
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001616811728914843,
+      "loss": 0.8465,
+      "step": 1145
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001616121734493668,
+      "loss": 0.8527,
+      "step": 1146
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016154312669343867,
+      "loss": 0.8608,
+      "step": 1147
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016147403267672305,
+      "loss": 0.2434,
+      "step": 1148
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016140489145227918,
+      "loss": 0.8675,
+      "step": 1149
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016133570307320274,
+      "loss": 0.8611,
+      "step": 1150
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016126646759262548,
+      "loss": 0.2613,
+      "step": 1151
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016119718506371535,
+      "loss": 0.8422,
+      "step": 1152
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001611278555396765,
+      "loss": 0.8758,
+      "step": 1153
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016105847907374912,
+      "loss": 0.8639,
+      "step": 1154
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016098905571920944,
+      "loss": 0.8848,
+      "step": 1155
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001609195855293697,
+      "loss": 0.8543,
+      "step": 1156
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001608500685575781,
+      "loss": 0.8284,
+      "step": 1157
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016078050485721882,
+      "loss": 0.8336,
+      "step": 1158
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016071089448171187,
+      "loss": 0.2341,
+      "step": 1159
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016064123748451312,
+      "loss": 0.8843,
+      "step": 1160
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016057153391911422,
+      "loss": 0.8568,
+      "step": 1161
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016050178383904257,
+      "loss": 0.8616,
+      "step": 1162
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001604319872978614,
+      "loss": 0.9044,
+      "step": 1163
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001603621443491695,
+      "loss": 0.8179,
+      "step": 1164
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001602922550466013,
+      "loss": 0.8543,
+      "step": 1165
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016022231944382693,
+      "loss": 0.8608,
+      "step": 1166
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016015233759455194,
+      "loss": 0.856,
+      "step": 1167
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016008230955251752,
+      "loss": 0.8665,
+      "step": 1168
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016001223537150024,
+      "loss": 0.8635,
+      "step": 1169
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015994211510531216,
+      "loss": 0.8729,
+      "step": 1170
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001598719488078007,
+      "loss": 0.833,
+      "step": 1171
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015980173653284862,
+      "loss": 0.8541,
+      "step": 1172
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015973147833437401,
+      "loss": 0.8297,
+      "step": 1173
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015966117426633022,
+      "loss": 0.8934,
+      "step": 1174
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015959082438270585,
+      "loss": 0.8796,
+      "step": 1175
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001595204287375246,
+      "loss": 0.8636,
+      "step": 1176
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015944998738484542,
+      "loss": 0.8426,
+      "step": 1177
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001593795003787623,
+      "loss": 0.8555,
+      "step": 1178
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015930896777340426,
+      "loss": 0.8061,
+      "step": 1179
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015923838962293542,
+      "loss": 0.8651,
+      "step": 1180
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015916776598155478,
+      "loss": 0.824,
+      "step": 1181
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001590970969034964,
+      "loss": 0.8866,
+      "step": 1182
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001590263824430291,
+      "loss": 0.8175,
+      "step": 1183
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015895562265445657,
+      "loss": 0.765,
+      "step": 1184
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001588848175921174,
+      "loss": 0.8447,
+      "step": 1185
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015881396731038492,
+      "loss": 0.2575,
+      "step": 1186
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001587430718636671,
+      "loss": 0.8376,
+      "step": 1187
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015867213130640667,
+      "loss": 0.8695,
+      "step": 1188
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015860114569308097,
+      "loss": 0.8571,
+      "step": 1189
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000158530115078202,
+      "loss": 0.8193,
+      "step": 1190
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015845903951631623,
+      "loss": 0.8337,
+      "step": 1191
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015838791906200466,
+      "loss": 0.858,
+      "step": 1192
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001583167537698828,
+      "loss": 0.869,
+      "step": 1193
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015824554369460068,
+      "loss": 0.8065,
+      "step": 1194
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001581742888908425,
+      "loss": 0.8494,
+      "step": 1195
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015810298941332698,
+      "loss": 0.8583,
+      "step": 1196
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015803164531680714,
+      "loss": 0.822,
+      "step": 1197
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015796025665607013,
+      "loss": 0.8281,
+      "step": 1198
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015788882348593747,
+      "loss": 0.8439,
+      "step": 1199
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015781734586126485,
+      "loss": 0.8079,
+      "step": 1200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015774582383694193,
+      "loss": 0.8207,
+      "step": 1201
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015767425746789272,
+      "loss": 0.7915,
+      "step": 1202
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015760264680907506,
+      "loss": 0.8073,
+      "step": 1203
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015753099191548095,
+      "loss": 0.8152,
+      "step": 1204
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015745929284213624,
+      "loss": 0.8205,
+      "step": 1205
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015738754964410084,
+      "loss": 0.2597,
+      "step": 1206
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001573157623764684,
+      "loss": 0.779,
+      "step": 1207
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015724393109436656,
+      "loss": 0.8657,
+      "step": 1208
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015717205585295666,
+      "loss": 0.8367,
+      "step": 1209
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001571001367074338,
+      "loss": 0.2486,
+      "step": 1210
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015702817371302685,
+      "loss": 0.8195,
+      "step": 1211
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015695616692499833,
+      "loss": 0.8622,
+      "step": 1212
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015688411639864431,
+      "loss": 0.8421,
+      "step": 1213
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015681202218929463,
+      "loss": 0.8624,
+      "step": 1214
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015673988435231247,
+      "loss": 0.8499,
+      "step": 1215
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015666770294309464,
+      "loss": 0.8557,
+      "step": 1216
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001565954780170714,
+      "loss": 0.8806,
+      "step": 1217
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001565232096297064,
+      "loss": 0.8461,
+      "step": 1218
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001564508978364967,
+      "loss": 0.8391,
+      "step": 1219
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015637854269297255,
+      "loss": 0.8307,
+      "step": 1220
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015630614425469775,
+      "loss": 0.8499,
+      "step": 1221
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015623370257726913,
+      "loss": 0.8494,
+      "step": 1222
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001561612177163168,
+      "loss": 0.8387,
+      "step": 1223
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015608868972750403,
+      "loss": 0.8743,
+      "step": 1224
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001560161186665272,
+      "loss": 0.817,
+      "step": 1225
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015594350458911585,
+      "loss": 0.8467,
+      "step": 1226
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015587084755103235,
+      "loss": 0.8644,
+      "step": 1227
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015579814760807226,
+      "loss": 0.8646,
+      "step": 1228
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015572540481606406,
+      "loss": 0.8629,
+      "step": 1229
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015565261923086902,
+      "loss": 0.8383,
+      "step": 1230
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015557979090838136,
+      "loss": 0.8174,
+      "step": 1231
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001555069199045281,
+      "loss": 0.8038,
+      "step": 1232
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015543400627526907,
+      "loss": 0.8411,
+      "step": 1233
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015536105007659678,
+      "loss": 0.82,
+      "step": 1234
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015528805136453647,
+      "loss": 0.7985,
+      "step": 1235
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015521501019514597,
+      "loss": 0.8057,
+      "step": 1236
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015514192662451578,
+      "loss": 0.8226,
+      "step": 1237
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015506880070876897,
+      "loss": 0.8814,
+      "step": 1238
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015499563250406104,
+      "loss": 0.8234,
+      "step": 1239
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015492242206658,
+      "loss": 0.9153,
+      "step": 1240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001548491694525464,
+      "loss": 0.8105,
+      "step": 1241
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015477587471821304,
+      "loss": 0.7915,
+      "step": 1242
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015470253791986514,
+      "loss": 0.8314,
+      "step": 1243
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015462915911382011,
+      "loss": 0.8522,
+      "step": 1244
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015455573835642777,
+      "loss": 0.8455,
+      "step": 1245
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015448227570407012,
+      "loss": 0.8426,
+      "step": 1246
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001544087712131613,
+      "loss": 0.8355,
+      "step": 1247
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015433522494014753,
+      "loss": 0.8856,
+      "step": 1248
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015426163694150718,
+      "loss": 0.8319,
+      "step": 1249
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015418800727375067,
+      "loss": 0.8753,
+      "step": 1250
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015411433599342038,
+      "loss": 0.8447,
+      "step": 1251
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015404062315709072,
+      "loss": 0.8526,
+      "step": 1252
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015396686882136791,
+      "loss": 0.8541,
+      "step": 1253
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015389307304289007,
+      "loss": 0.7988,
+      "step": 1254
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001538192358783272,
+      "loss": 0.2588,
+      "step": 1255
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015374535738438104,
+      "loss": 0.8188,
+      "step": 1256
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015367143761778503,
+      "loss": 0.8753,
+      "step": 1257
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015359747663530439,
+      "loss": 0.823,
+      "step": 1258
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015352347449373594,
+      "loss": 0.8724,
+      "step": 1259
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001534494312499081,
+      "loss": 0.9063,
+      "step": 1260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015337534696068086,
+      "loss": 0.2474,
+      "step": 1261
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001533012216829458,
+      "loss": 0.8957,
+      "step": 1262
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001532270554736258,
+      "loss": 0.8176,
+      "step": 1263
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001531528483896754,
+      "loss": 0.8719,
+      "step": 1264
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015307860048808034,
+      "loss": 0.7846,
+      "step": 1265
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015300431182585777,
+      "loss": 0.8157,
+      "step": 1266
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015292998246005623,
+      "loss": 0.8382,
+      "step": 1267
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015285561244775538,
+      "loss": 0.8988,
+      "step": 1268
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015278120184606617,
+      "loss": 0.8519,
+      "step": 1269
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015270675071213074,
+      "loss": 0.9058,
+      "step": 1270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015263225910312222,
+      "loss": 0.8312,
+      "step": 1271
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000152557727076245,
+      "loss": 0.8722,
+      "step": 1272
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001524831546887344,
+      "loss": 0.8695,
+      "step": 1273
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015240854199785673,
+      "loss": 0.8149,
+      "step": 1274
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015233388906090939,
+      "loss": 0.8387,
+      "step": 1275
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001522591959352205,
+      "loss": 0.8562,
+      "step": 1276
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001521844626781491,
+      "loss": 0.8458,
+      "step": 1277
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001521096893470851,
+      "loss": 0.9145,
+      "step": 1278
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001520348759994492,
+      "loss": 0.8121,
+      "step": 1279
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015196002269269267,
+      "loss": 0.8089,
+      "step": 1280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015188512948429765,
+      "loss": 0.855,
+      "step": 1281
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001518101964317769,
+      "loss": 0.2527,
+      "step": 1282
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001517352235926736,
+      "loss": 0.2424,
+      "step": 1283
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015166021102456172,
+      "loss": 0.8159,
+      "step": 1284
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001515851587850456,
+      "loss": 0.811,
+      "step": 1285
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015151006693176004,
+      "loss": 0.8364,
+      "step": 1286
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015143493552237033,
+      "loss": 0.8371,
+      "step": 1287
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015135976461457212,
+      "loss": 0.8444,
+      "step": 1288
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001512845542660913,
+      "loss": 0.8343,
+      "step": 1289
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001512093045346842,
+      "loss": 0.8353,
+      "step": 1290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001511340154781373,
+      "loss": 0.8346,
+      "step": 1291
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015105868715426727,
+      "loss": 0.8819,
+      "step": 1292
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015098331962092097,
+      "loss": 0.8758,
+      "step": 1293
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015090791293597534,
+      "loss": 0.8647,
+      "step": 1294
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015083246715733739,
+      "loss": 0.8243,
+      "step": 1295
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015075698234294423,
+      "loss": 0.78,
+      "step": 1296
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015068145855076285,
+      "loss": 0.8375,
+      "step": 1297
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001506058958387902,
+      "loss": 0.837,
+      "step": 1298
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001505302942650531,
+      "loss": 0.8421,
+      "step": 1299
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015045465388760826,
+      "loss": 0.8056,
+      "step": 1300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001503789747645422,
+      "loss": 0.7702,
+      "step": 1301
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015030325695397108,
+      "loss": 0.8406,
+      "step": 1302
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015022750051404087,
+      "loss": 0.8256,
+      "step": 1303
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015015170550292722,
+      "loss": 0.8512,
+      "step": 1304
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015007587197883533,
+      "loss": 0.8275,
+      "step": 1305
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 0.8185,
+      "step": 1306
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001499240896246856,
+      "loss": 0.7801,
+      "step": 1307
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014984814091118594,
+      "loss": 0.8755,
+      "step": 1308
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014977215391782423,
+      "loss": 0.805,
+      "step": 1309
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014969612870295321,
+      "loss": 0.8147,
+      "step": 1310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014962006532495488,
+      "loss": 0.8544,
+      "step": 1311
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001495439638422405,
+      "loss": 0.82,
+      "step": 1312
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001494678243132507,
+      "loss": 0.8953,
+      "step": 1313
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014939164679645528,
+      "loss": 0.8184,
+      "step": 1314
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014931543135035316,
+      "loss": 0.7949,
+      "step": 1315
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001492391780334725,
+      "loss": 0.7913,
+      "step": 1316
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001491628869043704,
+      "loss": 0.8045,
+      "step": 1317
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014908655802163315,
+      "loss": 0.8117,
+      "step": 1318
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014901019144387593,
+      "loss": 0.8646,
+      "step": 1319
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014893378722974284,
+      "loss": 0.8398,
+      "step": 1320
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014885734543790705,
+      "loss": 0.8449,
+      "step": 1321
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001487808661270704,
+      "loss": 0.8227,
+      "step": 1322
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001487043493559636,
+      "loss": 0.8841,
+      "step": 1323
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014862779518334617,
+      "loss": 0.7926,
+      "step": 1324
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014855120366800634,
+      "loss": 0.8633,
+      "step": 1325
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014847457486876097,
+      "loss": 0.9188,
+      "step": 1326
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001483979088444556,
+      "loss": 0.8511,
+      "step": 1327
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014832120565396432,
+      "loss": 0.809,
+      "step": 1328
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001482444653561898,
+      "loss": 0.2451,
+      "step": 1329
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014816768801006317,
+      "loss": 0.8668,
+      "step": 1330
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000148090873674544,
+      "loss": 0.8146,
+      "step": 1331
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014801402240862034,
+      "loss": 0.8275,
+      "step": 1332
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014793713427130857,
+      "loss": 0.8481,
+      "step": 1333
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001478602093216533,
+      "loss": 0.8189,
+      "step": 1334
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001477832476187275,
+      "loss": 0.816,
+      "step": 1335
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014770624922163232,
+      "loss": 0.8442,
+      "step": 1336
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014762921418949714,
+      "loss": 0.7936,
+      "step": 1337
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014755214258147943,
+      "loss": 0.8403,
+      "step": 1338
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014747503445676475,
+      "loss": 0.2551,
+      "step": 1339
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014739788987456673,
+      "loss": 0.8531,
+      "step": 1340
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014732070889412693,
+      "loss": 0.8483,
+      "step": 1341
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014724349157471496,
+      "loss": 0.8389,
+      "step": 1342
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014716623797562825,
+      "loss": 0.2432,
+      "step": 1343
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014708894815619212,
+      "loss": 0.8594,
+      "step": 1344
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014701162217575976,
+      "loss": 0.8127,
+      "step": 1345
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014693426009371202,
+      "loss": 0.8026,
+      "step": 1346
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001468568619694576,
+      "loss": 0.8405,
+      "step": 1347
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001467794278624327,
+      "loss": 0.8471,
+      "step": 1348
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014670195783210134,
+      "loss": 0.8473,
+      "step": 1349
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014662445193795506,
+      "loss": 0.8648,
+      "step": 1350
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001465469102395129,
+      "loss": 0.8514,
+      "step": 1351
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001464693327963214,
+      "loss": 0.8213,
+      "step": 1352
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001463917196679546,
+      "loss": 0.8525,
+      "step": 1353
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014631407091401393,
+      "loss": 0.8851,
+      "step": 1354
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014623638659412816,
+      "loss": 0.8228,
+      "step": 1355
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014615866676795334,
+      "loss": 0.8345,
+      "step": 1356
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001460809114951729,
+      "loss": 0.8261,
+      "step": 1357
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001460031208354973,
+      "loss": 0.2157,
+      "step": 1358
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001459252948486644,
+      "loss": 0.8566,
+      "step": 1359
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014584743359443903,
+      "loss": 0.8669,
+      "step": 1360
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014576953713261312,
+      "loss": 0.8153,
+      "step": 1361
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014569160552300574,
+      "loss": 0.2223,
+      "step": 1362
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014561363882546283,
+      "loss": 0.8695,
+      "step": 1363
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014553563709985732,
+      "loss": 0.8657,
+      "step": 1364
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014545760040608904,
+      "loss": 0.8375,
+      "step": 1365
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014537952880408473,
+      "loss": 0.8439,
+      "step": 1366
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014530142235379782,
+      "loss": 0.8304,
+      "step": 1367
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014522328111520857,
+      "loss": 0.8382,
+      "step": 1368
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014514510514832403,
+      "loss": 0.8326,
+      "step": 1369
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001450668945131777,
+      "loss": 0.8356,
+      "step": 1370
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014498864926982996,
+      "loss": 0.7849,
+      "step": 1371
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014491036947836757,
+      "loss": 0.2413,
+      "step": 1372
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014483205519890392,
+      "loss": 0.8712,
+      "step": 1373
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001447537064915789,
+      "loss": 0.8654,
+      "step": 1374
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014467532341655879,
+      "loss": 0.8217,
+      "step": 1375
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014459690603403622,
+      "loss": 0.2365,
+      "step": 1376
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014451845440423025,
+      "loss": 0.8563,
+      "step": 1377
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001444399685873862,
+      "loss": 0.8676,
+      "step": 1378
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014436144864377566,
+      "loss": 0.8093,
+      "step": 1379
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014428289463369644,
+      "loss": 0.8207,
+      "step": 1380
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014420430661747245,
+      "loss": 0.8355,
+      "step": 1381
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014412568465545374,
+      "loss": 0.8376,
+      "step": 1382
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014404702880801648,
+      "loss": 0.8257,
+      "step": 1383
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014396833913556283,
+      "loss": 0.7743,
+      "step": 1384
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014388961569852086,
+      "loss": 0.7934,
+      "step": 1385
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014381085855734468,
+      "loss": 0.7867,
+      "step": 1386
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001437320677725142,
+      "loss": 0.8619,
+      "step": 1387
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014365324340453523,
+      "loss": 0.8395,
+      "step": 1388
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014357438551393928,
+      "loss": 0.8162,
+      "step": 1389
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001434954941612837,
+      "loss": 0.2457,
+      "step": 1390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014341656940715147,
+      "loss": 0.8781,
+      "step": 1391
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014333761131215126,
+      "loss": 0.8204,
+      "step": 1392
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014325861993691734,
+      "loss": 0.8406,
+      "step": 1393
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014317959534210944,
+      "loss": 0.8316,
+      "step": 1394
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000143100537588413,
+      "loss": 0.8397,
+      "step": 1395
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014302144673653875,
+      "loss": 0.8429,
+      "step": 1396
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014294232284722287,
+      "loss": 0.8387,
+      "step": 1397
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000142863165981227,
+      "loss": 0.2665,
+      "step": 1398
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014278397619933798,
+      "loss": 0.8163,
+      "step": 1399
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000142704753562368,
+      "loss": 0.8158,
+      "step": 1400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001426254981311545,
+      "loss": 0.8044,
+      "step": 1401
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014254620996656002,
+      "loss": 0.8004,
+      "step": 1402
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014246688912947233,
+      "loss": 0.8329,
+      "step": 1403
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014238753568080422,
+      "loss": 0.2162,
+      "step": 1404
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001423081496814936,
+      "loss": 0.7888,
+      "step": 1405
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014222873119250324,
+      "loss": 0.7984,
+      "step": 1406
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000142149280274821,
+      "loss": 0.8167,
+      "step": 1407
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001420697969894596,
+      "loss": 0.8121,
+      "step": 1408
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001419902813974566,
+      "loss": 0.8546,
+      "step": 1409
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014191073355987436,
+      "loss": 0.8701,
+      "step": 1410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014183115353780002,
+      "loss": 0.8717,
+      "step": 1411
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014175154139234544,
+      "loss": 0.7932,
+      "step": 1412
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014167189718464714,
+      "loss": 0.8319,
+      "step": 1413
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014159222097586625,
+      "loss": 0.8252,
+      "step": 1414
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014151251282718853,
+      "loss": 0.8668,
+      "step": 1415
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014143277279982414,
+      "loss": 0.8601,
+      "step": 1416
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014135300095500792,
+      "loss": 0.8393,
+      "step": 1417
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014127319735399895,
+      "loss": 0.8306,
+      "step": 1418
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014119336205808077,
+      "loss": 0.8138,
+      "step": 1419
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001411134951285613,
+      "loss": 0.8209,
+      "step": 1420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014103359662677277,
+      "loss": 0.8146,
+      "step": 1421
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014095366661407148,
+      "loss": 0.819,
+      "step": 1422
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014087370515183812,
+      "loss": 0.8429,
+      "step": 1423
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014079371230147746,
+      "loss": 0.8289,
+      "step": 1424
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001407136881244184,
+      "loss": 0.8199,
+      "step": 1425
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001406336326821138,
+      "loss": 0.8486,
+      "step": 1426
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014055354603604068,
+      "loss": 0.8104,
+      "step": 1427
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001404734282476999,
+      "loss": 0.8281,
+      "step": 1428
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001403932793786163,
+      "loss": 0.8391,
+      "step": 1429
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014031309949033856,
+      "loss": 0.8576,
+      "step": 1430
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014023288864443916,
+      "loss": 0.8456,
+      "step": 1431
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001401526469025144,
+      "loss": 0.8706,
+      "step": 1432
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014007237432618434,
+      "loss": 0.8537,
+      "step": 1433
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001399920709770926,
+      "loss": 0.2495,
+      "step": 1434
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013991173691690656,
+      "loss": 0.8485,
+      "step": 1435
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000139831372207317,
+      "loss": 0.886,
+      "step": 1436
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013975097691003852,
+      "loss": 0.8166,
+      "step": 1437
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013967055108680891,
+      "loss": 0.839,
+      "step": 1438
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013959009479938962,
+      "loss": 0.8217,
+      "step": 1439
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013950960810956537,
+      "loss": 0.8128,
+      "step": 1440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001394290910791443,
+      "loss": 0.8334,
+      "step": 1441
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001393485437699578,
+      "loss": 0.8393,
+      "step": 1442
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013926796624386052,
+      "loss": 0.8673,
+      "step": 1443
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001391873585627304,
+      "loss": 0.8354,
+      "step": 1444
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013910672078846836,
+      "loss": 0.8306,
+      "step": 1445
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001390260529829986,
+      "loss": 0.8658,
+      "step": 1446
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013894535520826828,
+      "loss": 0.8438,
+      "step": 1447
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013886462752624762,
+      "loss": 0.2368,
+      "step": 1448
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001387838699989298,
+      "loss": 0.8098,
+      "step": 1449
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013870308268833093,
+      "loss": 0.8095,
+      "step": 1450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013862226565648995,
+      "loss": 0.8829,
+      "step": 1451
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013854141896546863,
+      "loss": 0.8333,
+      "step": 1452
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013846054267735157,
+      "loss": 0.8343,
+      "step": 1453
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013837963685424607,
+      "loss": 0.8354,
+      "step": 1454
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013829870155828209,
+      "loss": 0.8461,
+      "step": 1455
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013821773685161225,
+      "loss": 0.8186,
+      "step": 1456
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001381367427964117,
+      "loss": 0.8626,
+      "step": 1457
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001380557194548782,
+      "loss": 0.863,
+      "step": 1458
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000137974666889232,
+      "loss": 0.8389,
+      "step": 1459
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013789358516171573,
+      "loss": 0.8533,
+      "step": 1460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013781247433459449,
+      "loss": 0.8019,
+      "step": 1461
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013773133447015558,
+      "loss": 0.7943,
+      "step": 1462
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001376501656307088,
+      "loss": 0.8587,
+      "step": 1463
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013756896787858608,
+      "loss": 0.8151,
+      "step": 1464
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013748774127614154,
+      "loss": 0.8102,
+      "step": 1465
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013740648588575155,
+      "loss": 0.8102,
+      "step": 1466
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001373252017698145,
+      "loss": 0.8201,
+      "step": 1467
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013724388899075083,
+      "loss": 0.8507,
+      "step": 1468
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001371625476110031,
+      "loss": 0.8349,
+      "step": 1469
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001370811776930357,
+      "loss": 0.8647,
+      "step": 1470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013699977929933503,
+      "loss": 0.8249,
+      "step": 1471
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013691835249240932,
+      "loss": 0.8174,
+      "step": 1472
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001368368973347886,
+      "loss": 0.8585,
+      "step": 1473
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013675541388902477,
+      "loss": 0.8706,
+      "step": 1474
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013667390221769126,
+      "loss": 0.8466,
+      "step": 1475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001365923623833834,
+      "loss": 0.8348,
+      "step": 1476
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013651079444871794,
+      "loss": 0.8221,
+      "step": 1477
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001364291984763334,
+      "loss": 0.8401,
+      "step": 1478
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013634757452888973,
+      "loss": 0.8214,
+      "step": 1479
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013626592266906834,
+      "loss": 0.847,
+      "step": 1480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001361842429595721,
+      "loss": 0.8268,
+      "step": 1481
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013610253546312532,
+      "loss": 0.8147,
+      "step": 1482
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013602080024247357,
+      "loss": 0.8454,
+      "step": 1483
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013593903736038376,
+      "loss": 0.8555,
+      "step": 1484
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013585724687964403,
+      "loss": 0.7947,
+      "step": 1485
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013577542886306367,
+      "loss": 0.8481,
+      "step": 1486
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001356935833734732,
+      "loss": 0.2464,
+      "step": 1487
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013561171047372418,
+      "loss": 0.7939,
+      "step": 1488
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013552981022668922,
+      "loss": 0.8551,
+      "step": 1489
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013544788269526198,
+      "loss": 0.8425,
+      "step": 1490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013536592794235696,
+      "loss": 0.8165,
+      "step": 1491
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013528394603090974,
+      "loss": 0.8458,
+      "step": 1492
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013520193702387656,
+      "loss": 0.7948,
+      "step": 1493
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001351199009842346,
+      "loss": 0.8262,
+      "step": 1494
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013503783797498176,
+      "loss": 0.8383,
+      "step": 1495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001349557480591367,
+      "loss": 0.8709,
+      "step": 1496
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001348736312997386,
+      "loss": 0.8104,
+      "step": 1497
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013479148775984742,
+      "loss": 0.836,
+      "step": 1498
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001347093175025436,
+      "loss": 0.8254,
+      "step": 1499
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001346271205909281,
+      "loss": 0.793,
+      "step": 1500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013454489708812238,
+      "loss": 0.8379,
+      "step": 1501
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013446264705726827,
+      "loss": 0.2424,
+      "step": 1502
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013438037056152802,
+      "loss": 0.8417,
+      "step": 1503
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013429806766408417,
+      "loss": 0.8188,
+      "step": 1504
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001342157384281396,
+      "loss": 0.8428,
+      "step": 1505
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013413338291691726,
+      "loss": 0.8211,
+      "step": 1506
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001340510011936605,
+      "loss": 0.275,
+      "step": 1507
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013396859332163264,
+      "loss": 0.8299,
+      "step": 1508
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013388615936411712,
+      "loss": 0.8743,
+      "step": 1509
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013380369938441737,
+      "loss": 0.8224,
+      "step": 1510
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013372121344585694,
+      "loss": 0.8448,
+      "step": 1511
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013363870161177914,
+      "loss": 0.7852,
+      "step": 1512
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013355616394554728,
+      "loss": 0.8167,
+      "step": 1513
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013347360051054448,
+      "loss": 0.8584,
+      "step": 1514
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001333910113701736,
+      "loss": 0.7989,
+      "step": 1515
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001333083965878574,
+      "loss": 0.8422,
+      "step": 1516
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013322575622703807,
+      "loss": 0.8089,
+      "step": 1517
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013314309035117767,
+      "loss": 0.8184,
+      "step": 1518
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001330603990237578,
+      "loss": 0.8914,
+      "step": 1519
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013297768230827948,
+      "loss": 0.8567,
+      "step": 1520
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013289494026826336,
+      "loss": 0.8545,
+      "step": 1521
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001328121729672496,
+      "loss": 0.815,
+      "step": 1522
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001327293804687975,
+      "loss": 0.8242,
+      "step": 1523
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000132646562836486,
+      "loss": 0.8228,
+      "step": 1524
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013256372013391312,
+      "loss": 0.7999,
+      "step": 1525
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013248085242469626,
+      "loss": 0.8273,
+      "step": 1526
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013239795977247197,
+      "loss": 0.8772,
+      "step": 1527
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013231504224089598,
+      "loss": 0.2592,
+      "step": 1528
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013223209989364317,
+      "loss": 0.8115,
+      "step": 1529
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001321491327944073,
+      "loss": 0.8504,
+      "step": 1530
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013206614100690138,
+      "loss": 0.8641,
+      "step": 1531
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013198312459485715,
+      "loss": 0.8426,
+      "step": 1532
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001319000836220255,
+      "loss": 0.8237,
+      "step": 1533
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013181701815217592,
+      "loss": 0.8308,
+      "step": 1534
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001317339282490969,
+      "loss": 0.8404,
+      "step": 1535
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013165081397659563,
+      "loss": 0.8484,
+      "step": 1536
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013156767539849803,
+      "loss": 0.8942,
+      "step": 1537
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013148451257864868,
+      "loss": 0.8384,
+      "step": 1538
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001314013255809107,
+      "loss": 0.8301,
+      "step": 1539
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013131811446916592,
+      "loss": 0.8464,
+      "step": 1540
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013123487930731463,
+      "loss": 0.8554,
+      "step": 1541
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001311516201592755,
+      "loss": 0.8457,
+      "step": 1542
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013106833708898577,
+      "loss": 0.8072,
+      "step": 1543
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013098503016040086,
+      "loss": 0.8204,
+      "step": 1544
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013090169943749476,
+      "loss": 0.8506,
+      "step": 1545
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013081834498425952,
+      "loss": 0.8361,
+      "step": 1546
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013073496686470546,
+      "loss": 0.8291,
+      "step": 1547
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001306515651428612,
+      "loss": 0.8613,
+      "step": 1548
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013056813988277332,
+      "loss": 0.8688,
+      "step": 1549
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013048469114850656,
+      "loss": 0.7784,
+      "step": 1550
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001304012190041437,
+      "loss": 0.781,
+      "step": 1551
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013031772351378542,
+      "loss": 0.8778,
+      "step": 1552
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001302342047415504,
+      "loss": 0.8131,
+      "step": 1553
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013015066275157518,
+      "loss": 0.8211,
+      "step": 1554
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013006709760801413,
+      "loss": 0.836,
+      "step": 1555
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012998350937503937,
+      "loss": 0.8147,
+      "step": 1556
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001298998981168408,
+      "loss": 0.869,
+      "step": 1557
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000129816263897626,
+      "loss": 0.7983,
+      "step": 1558
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001297326067816201,
+      "loss": 0.8309,
+      "step": 1559
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012964892683306592,
+      "loss": 0.8749,
+      "step": 1560
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012956522411622376,
+      "loss": 0.836,
+      "step": 1561
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001294814986953714,
+      "loss": 0.8713,
+      "step": 1562
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001293977506348041,
+      "loss": 0.8368,
+      "step": 1563
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001293139799988344,
+      "loss": 0.817,
+      "step": 1564
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012923018685179235,
+      "loss": 0.8273,
+      "step": 1565
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012914637125802512,
+      "loss": 0.8524,
+      "step": 1566
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001290625332818972,
+      "loss": 0.8161,
+      "step": 1567
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012897867298779025,
+      "loss": 0.8851,
+      "step": 1568
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012889479044010308,
+      "loss": 0.8584,
+      "step": 1569
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012881088570325155,
+      "loss": 0.8206,
+      "step": 1570
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012872695884166862,
+      "loss": 0.8087,
+      "step": 1571
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001286430099198042,
+      "loss": 0.8799,
+      "step": 1572
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012855903900212514,
+      "loss": 0.8076,
+      "step": 1573
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001284750461531152,
+      "loss": 0.7617,
+      "step": 1574
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000128391031437275,
+      "loss": 0.8084,
+      "step": 1575
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012830699491912185,
+      "loss": 0.8551,
+      "step": 1576
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012822293666319,
+      "loss": 0.8331,
+      "step": 1577
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012813885673403015,
+      "loss": 0.8207,
+      "step": 1578
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012805475519620983,
+      "loss": 0.8214,
+      "step": 1579
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012797063211431313,
+      "loss": 0.2584,
+      "step": 1580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012788648755294055,
+      "loss": 0.7965,
+      "step": 1581
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012780232157670924,
+      "loss": 0.8327,
+      "step": 1582
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001277181342502528,
+      "loss": 0.8501,
+      "step": 1583
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012763392563822106,
+      "loss": 0.8618,
+      "step": 1584
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001275496958052804,
+      "loss": 0.827,
+      "step": 1585
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012746544481611336,
+      "loss": 0.8256,
+      "step": 1586
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012738117273541875,
+      "loss": 0.2589,
+      "step": 1587
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001272968796279116,
+      "loss": 0.865,
+      "step": 1588
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012721256555832316,
+      "loss": 0.8669,
+      "step": 1589
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012712823059140055,
+      "loss": 0.8195,
+      "step": 1590
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001270438747919072,
+      "loss": 0.7838,
+      "step": 1591
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012695949822462242,
+      "loss": 0.8484,
+      "step": 1592
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012687510095434138,
+      "loss": 0.2598,
+      "step": 1593
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012679068304587534,
+      "loss": 0.8369,
+      "step": 1594
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012670624456405128,
+      "loss": 0.8352,
+      "step": 1595
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012662178557371198,
+      "loss": 0.7912,
+      "step": 1596
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012653730613971606,
+      "loss": 0.8083,
+      "step": 1597
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001264528063269377,
+      "loss": 0.8245,
+      "step": 1598
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012636828620026682,
+      "loss": 0.8242,
+      "step": 1599
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012628374582460898,
+      "loss": 0.7946,
+      "step": 1600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012619918526488518,
+      "loss": 0.8346,
+      "step": 1601
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000126114604586032,
+      "loss": 0.7983,
+      "step": 1602
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012603000385300143,
+      "loss": 0.8231,
+      "step": 1603
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012594538313076092,
+      "loss": 0.2266,
+      "step": 1604
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012586074248429317,
+      "loss": 0.8427,
+      "step": 1605
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012577608197859627,
+      "loss": 0.8473,
+      "step": 1606
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001256914016786835,
+      "loss": 0.8745,
+      "step": 1607
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012560670164958338,
+      "loss": 0.8183,
+      "step": 1608
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001255219819563396,
+      "loss": 0.2694,
+      "step": 1609
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012543724266401087,
+      "loss": 0.7951,
+      "step": 1610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000125352483837671,
+      "loss": 0.8475,
+      "step": 1611
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012526770554240887,
+      "loss": 0.8189,
+      "step": 1612
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012518290784332812,
+      "loss": 0.8114,
+      "step": 1613
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012509809080554754,
+      "loss": 0.8159,
+      "step": 1614
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012501325449420055,
+      "loss": 0.817,
+      "step": 1615
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012492839897443553,
+      "loss": 0.8301,
+      "step": 1616
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012484352431141552,
+      "loss": 0.8137,
+      "step": 1617
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012475863057031825,
+      "loss": 0.8196,
+      "step": 1618
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012467371781633613,
+      "loss": 0.8392,
+      "step": 1619
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012458878611467622,
+      "loss": 0.8033,
+      "step": 1620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012450383553056011,
+      "loss": 0.857,
+      "step": 1621
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001244188661292238,
+      "loss": 0.2704,
+      "step": 1622
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012433387797591782,
+      "loss": 0.8253,
+      "step": 1623
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012424887113590712,
+      "loss": 0.8343,
+      "step": 1624
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012416384567447096,
+      "loss": 0.8694,
+      "step": 1625
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012407880165690287,
+      "loss": 0.8325,
+      "step": 1626
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001239937391485107,
+      "loss": 0.8541,
+      "step": 1627
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001239086582146165,
+      "loss": 0.8435,
+      "step": 1628
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012382355892055636,
+      "loss": 0.836,
+      "step": 1629
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001237384413316806,
+      "loss": 0.8705,
+      "step": 1630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012365330551335348,
+      "loss": 0.8249,
+      "step": 1631
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012356815153095337,
+      "loss": 0.8646,
+      "step": 1632
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012348297944987253,
+      "loss": 0.8638,
+      "step": 1633
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012339778933551706,
+      "loss": 0.8269,
+      "step": 1634
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012331258125330706,
+      "loss": 0.8458,
+      "step": 1635
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012322735526867624,
+      "loss": 0.8463,
+      "step": 1636
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012314211144707221,
+      "loss": 0.8428,
+      "step": 1637
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012305684985395625,
+      "loss": 0.2623,
+      "step": 1638
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001229715705548032,
+      "loss": 0.8125,
+      "step": 1639
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012288627361510158,
+      "loss": 0.7987,
+      "step": 1640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012280095910035342,
+      "loss": 0.2345,
+      "step": 1641
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001227156270760743,
+      "loss": 0.8218,
+      "step": 1642
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012263027760779318,
+      "loss": 0.8209,
+      "step": 1643
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012254491076105242,
+      "loss": 0.797,
+      "step": 1644
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012245952660140777,
+      "loss": 0.8617,
+      "step": 1645
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012237412519442827,
+      "loss": 0.8556,
+      "step": 1646
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012228870660569615,
+      "loss": 0.8288,
+      "step": 1647
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012220327090080687,
+      "loss": 0.8281,
+      "step": 1648
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001221178181453691,
+      "loss": 0.8911,
+      "step": 1649
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012203234840500446,
+      "loss": 0.8264,
+      "step": 1650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001219468617453477,
+      "loss": 0.8518,
+      "step": 1651
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012186135823204661,
+      "loss": 0.7803,
+      "step": 1652
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012177583793076184,
+      "loss": 0.2474,
+      "step": 1653
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012169030090716694,
+      "loss": 0.8799,
+      "step": 1654
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012160474722694835,
+      "loss": 0.876,
+      "step": 1655
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012151917695580523,
+      "loss": 0.844,
+      "step": 1656
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012143359015944958,
+      "loss": 0.8175,
+      "step": 1657
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012134798690360599,
+      "loss": 0.8686,
+      "step": 1658
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012126236725401173,
+      "loss": 0.8469,
+      "step": 1659
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012117673127641669,
+      "loss": 0.8614,
+      "step": 1660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012109107903658325,
+      "loss": 0.8416,
+      "step": 1661
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012100541060028631,
+      "loss": 0.8172,
+      "step": 1662
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012091972603331318,
+      "loss": 0.8008,
+      "step": 1663
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012083402540146365,
+      "loss": 0.8318,
+      "step": 1664
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012074830877054963,
+      "loss": 0.8613,
+      "step": 1665
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012066257620639556,
+      "loss": 0.8003,
+      "step": 1666
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012057682777483799,
+      "loss": 0.8471,
+      "step": 1667
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012049106354172566,
+      "loss": 0.796,
+      "step": 1668
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012040528357291947,
+      "loss": 0.8208,
+      "step": 1669
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012031948793429241,
+      "loss": 0.8486,
+      "step": 1670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012023367669172946,
+      "loss": 0.2697,
+      "step": 1671
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012014784991112767,
+      "loss": 0.7574,
+      "step": 1672
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012006200765839591,
+      "loss": 0.8452,
+      "step": 1673
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011997614999945499,
+      "loss": 0.8447,
+      "step": 1674
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011989027700023753,
+      "loss": 0.7819,
+      "step": 1675
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011980438872668803,
+      "loss": 0.8122,
+      "step": 1676
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001197184852447625,
+      "loss": 0.8704,
+      "step": 1677
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001196325666204289,
+      "loss": 0.8565,
+      "step": 1678
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011954663291966661,
+      "loss": 0.854,
+      "step": 1679
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011946068420846666,
+      "loss": 0.8253,
+      "step": 1680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011937472055283167,
+      "loss": 0.8103,
+      "step": 1681
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011928874201877558,
+      "loss": 0.7714,
+      "step": 1682
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001192027486723239,
+      "loss": 0.7782,
+      "step": 1683
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001191167405795135,
+      "loss": 0.902,
+      "step": 1684
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011903071780639248,
+      "loss": 0.8763,
+      "step": 1685
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001189446804190203,
+      "loss": 0.8301,
+      "step": 1686
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001188586284834676,
+      "loss": 0.8786,
+      "step": 1687
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011877256206581619,
+      "loss": 0.8509,
+      "step": 1688
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011868648123215908,
+      "loss": 0.8086,
+      "step": 1689
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011860038604860024,
+      "loss": 0.2298,
+      "step": 1690
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011851427658125474,
+      "loss": 0.8226,
+      "step": 1691
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011842815289624853,
+      "loss": 0.8159,
+      "step": 1692
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011834201505971862,
+      "loss": 0.7916,
+      "step": 1693
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011825586313781271,
+      "loss": 0.7881,
+      "step": 1694
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011816969719668949,
+      "loss": 0.8516,
+      "step": 1695
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001180835173025183,
+      "loss": 0.8041,
+      "step": 1696
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011799732352147918,
+      "loss": 0.8294,
+      "step": 1697
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011791111591976298,
+      "loss": 0.8365,
+      "step": 1698
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011782489456357102,
+      "loss": 0.2554,
+      "step": 1699
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001177386595191152,
+      "loss": 0.8366,
+      "step": 1700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011765241085261802,
+      "loss": 0.8431,
+      "step": 1701
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011756614863031237,
+      "loss": 0.7917,
+      "step": 1702
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011747987291844154,
+      "loss": 0.8908,
+      "step": 1703
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001173935837832592,
+      "loss": 0.8156,
+      "step": 1704
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011730728129102932,
+      "loss": 0.826,
+      "step": 1705
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001172209655080262,
+      "loss": 0.8337,
+      "step": 1706
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011713463650053423,
+      "loss": 0.8711,
+      "step": 1707
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000117048294334848,
+      "loss": 0.7671,
+      "step": 1708
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011696193907727222,
+      "loss": 0.8165,
+      "step": 1709
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011687557079412171,
+      "loss": 0.8501,
+      "step": 1710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011678918955172112,
+      "loss": 0.8512,
+      "step": 1711
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011670279541640521,
+      "loss": 0.8049,
+      "step": 1712
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011661638845451866,
+      "loss": 0.86,
+      "step": 1713
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011652996873241581,
+      "loss": 0.8507,
+      "step": 1714
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011644353631646103,
+      "loss": 0.8579,
+      "step": 1715
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011635709127302829,
+      "loss": 0.834,
+      "step": 1716
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011627063366850124,
+      "loss": 0.853,
+      "step": 1717
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011618416356927333,
+      "loss": 0.8067,
+      "step": 1718
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011609768104174743,
+      "loss": 0.834,
+      "step": 1719
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011601118615233608,
+      "loss": 0.8488,
+      "step": 1720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011592467896746122,
+      "loss": 0.8238,
+      "step": 1721
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011583815955355435,
+      "loss": 0.8194,
+      "step": 1722
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011575162797705616,
+      "loss": 0.8477,
+      "step": 1723
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011566508430441693,
+      "loss": 0.8829,
+      "step": 1724
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011557852860209608,
+      "loss": 0.83,
+      "step": 1725
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011549196093656222,
+      "loss": 0.2303,
+      "step": 1726
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011540538137429334,
+      "loss": 0.827,
+      "step": 1727
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011531878998177635,
+      "loss": 0.8466,
+      "step": 1728
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011523218682550736,
+      "loss": 0.2493,
+      "step": 1729
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011514557197199152,
+      "loss": 0.817,
+      "step": 1730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011505894548774294,
+      "loss": 0.8473,
+      "step": 1731
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011497230743928467,
+      "loss": 0.7917,
+      "step": 1732
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011488565789314858,
+      "loss": 0.2587,
+      "step": 1733
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011479899691587546,
+      "loss": 0.7794,
+      "step": 1734
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011471232457401482,
+      "loss": 0.8311,
+      "step": 1735
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011462564093412493,
+      "loss": 0.82,
+      "step": 1736
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011453894606277271,
+      "loss": 0.891,
+      "step": 1737
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011445224002653369,
+      "loss": 0.8,
+      "step": 1738
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001143655228919921,
+      "loss": 0.8124,
+      "step": 1739
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011427879472574048,
+      "loss": 0.7883,
+      "step": 1740
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011419205559437998,
+      "loss": 0.8313,
+      "step": 1741
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011410530556452021,
+      "loss": 0.8363,
+      "step": 1742
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000114018544702779,
+      "loss": 0.7961,
+      "step": 1743
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011393177307578262,
+      "loss": 0.8306,
+      "step": 1744
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011384499075016559,
+      "loss": 0.8205,
+      "step": 1745
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011375819779257057,
+      "loss": 0.2564,
+      "step": 1746
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011367139426964851,
+      "loss": 0.821,
+      "step": 1747
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011358458024805833,
+      "loss": 0.8269,
+      "step": 1748
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011349775579446716,
+      "loss": 0.8768,
+      "step": 1749
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011341092097555,
+      "loss": 0.8148,
+      "step": 1750
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011332407585798992,
+      "loss": 0.8384,
+      "step": 1751
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011323722050847782,
+      "loss": 0.2626,
+      "step": 1752
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011315035499371254,
+      "loss": 0.8483,
+      "step": 1753
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011306347938040062,
+      "loss": 0.8465,
+      "step": 1754
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011297659373525645,
+      "loss": 0.8336,
+      "step": 1755
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011288969812500207,
+      "loss": 0.7932,
+      "step": 1756
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011280279261636722,
+      "loss": 0.8611,
+      "step": 1757
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011271587727608917,
+      "loss": 0.8624,
+      "step": 1758
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011262895217091283,
+      "loss": 0.7919,
+      "step": 1759
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011254201736759051,
+      "loss": 0.8279,
+      "step": 1760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011245507293288204,
+      "loss": 0.8742,
+      "step": 1761
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011236811893355463,
+      "loss": 0.8098,
+      "step": 1762
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001122811554363828,
+      "loss": 0.8257,
+      "step": 1763
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011219418250814841,
+      "loss": 0.8481,
+      "step": 1764
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011210720021564058,
+      "loss": 0.7808,
+      "step": 1765
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011202020862565554,
+      "loss": 0.8342,
+      "step": 1766
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001119332078049967,
+      "loss": 0.7962,
+      "step": 1767
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011184619782047461,
+      "loss": 0.8067,
+      "step": 1768
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011175917873890673,
+      "loss": 0.7641,
+      "step": 1769
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011167215062711767,
+      "loss": 0.772,
+      "step": 1770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011158511355193888,
+      "loss": 0.8362,
+      "step": 1771
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011149806758020864,
+      "loss": 0.8506,
+      "step": 1772
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011141101277877222,
+      "loss": 0.804,
+      "step": 1773
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001113239492144815,
+      "loss": 0.8175,
+      "step": 1774
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011123687695419518,
+      "loss": 0.8441,
+      "step": 1775
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011114979606477866,
+      "loss": 0.8246,
+      "step": 1776
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011106270661310391,
+      "loss": 0.8447,
+      "step": 1777
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011097560866604953,
+      "loss": 0.8532,
+      "step": 1778
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011088850229050052,
+      "loss": 0.8268,
+      "step": 1779
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011080138755334853,
+      "loss": 0.8152,
+      "step": 1780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011071426452149151,
+      "loss": 0.8187,
+      "step": 1781
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011062713326183383,
+      "loss": 0.7885,
+      "step": 1782
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011053999384128613,
+      "loss": 0.8661,
+      "step": 1783
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011045284632676536,
+      "loss": 0.8318,
+      "step": 1784
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001103656907851947,
+      "loss": 0.8466,
+      "step": 1785
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011027852728350343,
+      "loss": 0.8051,
+      "step": 1786
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011019135588862702,
+      "loss": 0.8279,
+      "step": 1787
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011010417666750695,
+      "loss": 0.7935,
+      "step": 1788
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011001698968709069,
+      "loss": 0.7975,
+      "step": 1789
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010992979501433174,
+      "loss": 0.2416,
+      "step": 1790
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010984259271618947,
+      "loss": 0.8308,
+      "step": 1791
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010975538285962907,
+      "loss": 0.8562,
+      "step": 1792
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001096681655116216,
+      "loss": 0.859,
+      "step": 1793
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010958094073914384,
+      "loss": 0.8162,
+      "step": 1794
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010949370860917827,
+      "loss": 0.864,
+      "step": 1795
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000109406469188713,
+      "loss": 0.8027,
+      "step": 1796
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001093192225447418,
+      "loss": 0.8291,
+      "step": 1797
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010923196874426391,
+      "loss": 0.781,
+      "step": 1798
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010914470785428412,
+      "loss": 0.8715,
+      "step": 1799
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010905743994181267,
+      "loss": 0.8449,
+      "step": 1800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010897016507386512,
+      "loss": 0.7685,
+      "step": 1801
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001088828833174625,
+      "loss": 0.8275,
+      "step": 1802
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010879559473963099,
+      "loss": 0.8033,
+      "step": 1803
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010870829940740209,
+      "loss": 0.8096,
+      "step": 1804
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010862099738781246,
+      "loss": 0.2468,
+      "step": 1805
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010853368874790392,
+      "loss": 0.2668,
+      "step": 1806
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010844637355472335,
+      "loss": 0.7801,
+      "step": 1807
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010835905187532267,
+      "loss": 0.8152,
+      "step": 1808
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010827172377675876,
+      "loss": 0.8563,
+      "step": 1809
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010818438932609349,
+      "loss": 0.7972,
+      "step": 1810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010809704859039355,
+      "loss": 0.7922,
+      "step": 1811
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010800970163673048,
+      "loss": 0.8225,
+      "step": 1812
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010792234853218053,
+      "loss": 0.8183,
+      "step": 1813
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010783498934382484,
+      "loss": 0.8572,
+      "step": 1814
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010774762413874904,
+      "loss": 0.8278,
+      "step": 1815
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010766025298404346,
+      "loss": 0.8185,
+      "step": 1816
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010757287594680305,
+      "loss": 0.8549,
+      "step": 1817
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010748549309412714,
+      "loss": 0.8854,
+      "step": 1818
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010739810449311967,
+      "loss": 0.8377,
+      "step": 1819
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010731071021088889,
+      "loss": 0.8329,
+      "step": 1820
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010722331031454748,
+      "loss": 0.8083,
+      "step": 1821
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010713590487121243,
+      "loss": 0.8293,
+      "step": 1822
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010704849394800492,
+      "loss": 0.7992,
+      "step": 1823
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010696107761205038,
+      "loss": 0.7712,
+      "step": 1824
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010687365593047845,
+      "loss": 0.8761,
+      "step": 1825
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010678622897042279,
+      "loss": 0.8232,
+      "step": 1826
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010669879679902115,
+      "loss": 0.7977,
+      "step": 1827
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010661135948341526,
+      "loss": 0.8127,
+      "step": 1828
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010652391709075088,
+      "loss": 0.8478,
+      "step": 1829
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010643646968817757,
+      "loss": 0.8309,
+      "step": 1830
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010634901734284879,
+      "loss": 0.7875,
+      "step": 1831
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010626156012192176,
+      "loss": 0.8631,
+      "step": 1832
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010617409809255748,
+      "loss": 0.8315,
+      "step": 1833
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010608663132192064,
+      "loss": 0.7705,
+      "step": 1834
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010599915987717958,
+      "loss": 0.8447,
+      "step": 1835
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010591168382550616,
+      "loss": 0.2303,
+      "step": 1836
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010582420323407587,
+      "loss": 0.824,
+      "step": 1837
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010573671817006761,
+      "loss": 0.8125,
+      "step": 1838
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010564922870066376,
+      "loss": 0.8124,
+      "step": 1839
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010556173489305007,
+      "loss": 0.851,
+      "step": 1840
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010547423681441567,
+      "loss": 0.8331,
+      "step": 1841
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010538673453195285,
+      "loss": 0.7943,
+      "step": 1842
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010529922811285724,
+      "loss": 0.8064,
+      "step": 1843
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001052117176243276,
+      "loss": 0.8699,
+      "step": 1844
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010512420313356581,
+      "loss": 0.8473,
+      "step": 1845
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001050366847077769,
+      "loss": 0.8477,
+      "step": 1846
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010494916241416874,
+      "loss": 0.8146,
+      "step": 1847
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001048616363199524,
+      "loss": 0.8139,
+      "step": 1848
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010477410649234172,
+      "loss": 0.8541,
+      "step": 1849
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001046865729985534,
+      "loss": 0.8184,
+      "step": 1850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010459903590580706,
+      "loss": 0.8488,
+      "step": 1851
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010451149528132503,
+      "loss": 0.8303,
+      "step": 1852
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010442395119233228,
+      "loss": 0.8041,
+      "step": 1853
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010433640370605652,
+      "loss": 0.8517,
+      "step": 1854
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010424885288972812,
+      "loss": 0.8045,
+      "step": 1855
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010416129881057987,
+      "loss": 0.8268,
+      "step": 1856
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010407374153584716,
+      "loss": 0.7992,
+      "step": 1857
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010398618113276786,
+      "loss": 0.8153,
+      "step": 1858
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010389861766858211,
+      "loss": 0.8274,
+      "step": 1859
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010381105121053255,
+      "loss": 0.8528,
+      "step": 1860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000103723481825864,
+      "loss": 0.839,
+      "step": 1861
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001036359095818236,
+      "loss": 0.8327,
+      "step": 1862
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010354833454566071,
+      "loss": 0.8323,
+      "step": 1863
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010346075678462675,
+      "loss": 0.8336,
+      "step": 1864
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010337317636597528,
+      "loss": 0.8138,
+      "step": 1865
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010328559335696188,
+      "loss": 0.826,
+      "step": 1866
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010319800782484415,
+      "loss": 0.828,
+      "step": 1867
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010311041983688162,
+      "loss": 0.8042,
+      "step": 1868
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010302282946033565,
+      "loss": 0.8588,
+      "step": 1869
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010293523676246956,
+      "loss": 0.8071,
+      "step": 1870
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010284764181054831,
+      "loss": 0.7928,
+      "step": 1871
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010276004467183868,
+      "loss": 0.8553,
+      "step": 1872
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001026724454136091,
+      "loss": 0.7939,
+      "step": 1873
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001025848441031296,
+      "loss": 0.8018,
+      "step": 1874
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010249724080767192,
+      "loss": 0.8467,
+      "step": 1875
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010240963559450909,
+      "loss": 0.8432,
+      "step": 1876
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010232202853091583,
+      "loss": 0.8144,
+      "step": 1877
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010223441968416818,
+      "loss": 0.8069,
+      "step": 1878
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010214680912154354,
+      "loss": 0.8381,
+      "step": 1879
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010205919691032067,
+      "loss": 0.8459,
+      "step": 1880
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010197158311777957,
+      "loss": 0.224,
+      "step": 1881
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010188396781120149,
+      "loss": 0.8331,
+      "step": 1882
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010179635105786878,
+      "loss": 0.8019,
+      "step": 1883
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010170873292506495,
+      "loss": 0.8481,
+      "step": 1884
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010162111348007459,
+      "loss": 0.8404,
+      "step": 1885
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001015334927901832,
+      "loss": 0.7767,
+      "step": 1886
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010144587092267736,
+      "loss": 0.7635,
+      "step": 1887
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010135824794484443,
+      "loss": 0.7794,
+      "step": 1888
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010127062392397277,
+      "loss": 0.8184,
+      "step": 1889
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001011829989273514,
+      "loss": 0.8086,
+      "step": 1890
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010109537302227013,
+      "loss": 0.8315,
+      "step": 1891
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010100774627601953,
+      "loss": 0.7914,
+      "step": 1892
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010092011875589076,
+      "loss": 0.8,
+      "step": 1893
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010083249052917556,
+      "loss": 0.8268,
+      "step": 1894
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010074486166316628,
+      "loss": 0.7878,
+      "step": 1895
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010065723222515566,
+      "loss": 0.827,
+      "step": 1896
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010056960228243701,
+      "loss": 0.8332,
+      "step": 1897
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001004819719023039,
+      "loss": 0.8597,
+      "step": 1898
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010039434115205034,
+      "loss": 0.8067,
+      "step": 1899
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010030671009897048,
+      "loss": 0.8499,
+      "step": 1900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001002190788103589,
+      "loss": 0.8052,
+      "step": 1901
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010013144735351023,
+      "loss": 0.7906,
+      "step": 1902
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010004381579571922,
+      "loss": 0.7905,
+      "step": 1903
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99561842042808e-05,
+      "loss": 0.8315,
+      "step": 1904
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.986855264648982e-05,
+      "loss": 0.8381,
+      "step": 1905
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97809211896411e-05,
+      "loss": 0.8153,
+      "step": 1906
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.969328990102954e-05,
+      "loss": 0.8281,
+      "step": 1907
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.960565884794973e-05,
+      "loss": 0.8265,
+      "step": 1908
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.951802809769612e-05,
+      "loss": 0.8174,
+      "step": 1909
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.943039771756301e-05,
+      "loss": 0.821,
+      "step": 1910
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.934276777484436e-05,
+      "loss": 0.8099,
+      "step": 1911
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.925513833683377e-05,
+      "loss": 0.801,
+      "step": 1912
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.916750947082445e-05,
+      "loss": 0.7955,
+      "step": 1913
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.907988124410928e-05,
+      "loss": 0.8104,
+      "step": 1914
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.89922537239805e-05,
+      "loss": 0.836,
+      "step": 1915
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.890462697772988e-05,
+      "loss": 0.8309,
+      "step": 1916
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.881700107264863e-05,
+      "loss": 0.8035,
+      "step": 1917
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.872937607602726e-05,
+      "loss": 0.8109,
+      "step": 1918
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.864175205515556e-05,
+      "loss": 0.8436,
+      "step": 1919
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.855412907732266e-05,
+      "loss": 0.8553,
+      "step": 1920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.84665072098168e-05,
+      "loss": 0.8027,
+      "step": 1921
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.837888651992546e-05,
+      "loss": 0.8178,
+      "step": 1922
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.829126707493504e-05,
+      "loss": 0.8412,
+      "step": 1923
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.820364894213125e-05,
+      "loss": 0.8394,
+      "step": 1924
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.811603218879856e-05,
+      "loss": 0.8376,
+      "step": 1925
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.802841688222043e-05,
+      "loss": 0.2219,
+      "step": 1926
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.794080308967935e-05,
+      "loss": 0.7604,
+      "step": 1927
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.78531908784565e-05,
+      "loss": 0.8164,
+      "step": 1928
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.776558031583186e-05,
+      "loss": 0.8184,
+      "step": 1929
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.767797146908418e-05,
+      "loss": 0.9128,
+      "step": 1930
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.759036440549094e-05,
+      "loss": 0.8245,
+      "step": 1931
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.750275919232812e-05,
+      "loss": 0.8353,
+      "step": 1932
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.741515589687038e-05,
+      "loss": 0.8146,
+      "step": 1933
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.732755458639093e-05,
+      "loss": 0.8362,
+      "step": 1934
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.723995532816134e-05,
+      "loss": 0.8293,
+      "step": 1935
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.71523581894517e-05,
+      "loss": 0.8172,
+      "step": 1936
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.706476323753046e-05,
+      "loss": 0.776,
+      "step": 1937
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.697717053966436e-05,
+      "loss": 0.7875,
+      "step": 1938
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.688958016311845e-05,
+      "loss": 0.7857,
+      "step": 1939
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.680199217515586e-05,
+      "loss": 0.812,
+      "step": 1940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.671440664303814e-05,
+      "loss": 0.8473,
+      "step": 1941
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.662682363402477e-05,
+      "loss": 0.7761,
+      "step": 1942
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.653924321537325e-05,
+      "loss": 0.8288,
+      "step": 1943
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.645166545433931e-05,
+      "loss": 0.8148,
+      "step": 1944
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.63640904181764e-05,
+      "loss": 0.8725,
+      "step": 1945
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.627651817413604e-05,
+      "loss": 0.7984,
+      "step": 1946
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.618894878946748e-05,
+      "loss": 0.8167,
+      "step": 1947
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.610138233141792e-05,
+      "loss": 0.842,
+      "step": 1948
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.601381886723219e-05,
+      "loss": 0.8301,
+      "step": 1949
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.592625846415283e-05,
+      "loss": 0.8174,
+      "step": 1950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.583870118942014e-05,
+      "loss": 0.8028,
+      "step": 1951
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.57511471102719e-05,
+      "loss": 0.7978,
+      "step": 1952
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.566359629394348e-05,
+      "loss": 0.7814,
+      "step": 1953
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.557604880766775e-05,
+      "loss": 0.8077,
+      "step": 1954
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.548850471867501e-05,
+      "loss": 0.254,
+      "step": 1955
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.540096409419296e-05,
+      "loss": 0.805,
+      "step": 1956
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.53134270014466e-05,
+      "loss": 0.817,
+      "step": 1957
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.522589350765831e-05,
+      "loss": 0.8215,
+      "step": 1958
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.513836368004763e-05,
+      "loss": 0.8249,
+      "step": 1959
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.505083758583125e-05,
+      "loss": 0.8232,
+      "step": 1960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.496331529222313e-05,
+      "loss": 0.7904,
+      "step": 1961
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.487579686643422e-05,
+      "loss": 0.7734,
+      "step": 1962
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.478828237567245e-05,
+      "loss": 0.8217,
+      "step": 1963
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.47007718871428e-05,
+      "loss": 0.2509,
+      "step": 1964
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.461326546804719e-05,
+      "loss": 0.8048,
+      "step": 1965
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.452576318558437e-05,
+      "loss": 0.8062,
+      "step": 1966
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.443826510694994e-05,
+      "loss": 0.8513,
+      "step": 1967
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.435077129933627e-05,
+      "loss": 0.8169,
+      "step": 1968
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.426328182993243e-05,
+      "loss": 0.7903,
+      "step": 1969
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.417579676592415e-05,
+      "loss": 0.8458,
+      "step": 1970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.408831617449385e-05,
+      "loss": 0.8319,
+      "step": 1971
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.400084012282044e-05,
+      "loss": 0.8517,
+      "step": 1972
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.391336867807938e-05,
+      "loss": 0.8096,
+      "step": 1973
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.382590190744252e-05,
+      "loss": 0.7687,
+      "step": 1974
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.373843987807826e-05,
+      "loss": 0.7911,
+      "step": 1975
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.365098265715125e-05,
+      "loss": 0.8401,
+      "step": 1976
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.356353031182244e-05,
+      "loss": 0.8323,
+      "step": 1977
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.347608290924913e-05,
+      "loss": 0.8227,
+      "step": 1978
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.338864051658477e-05,
+      "loss": 0.8552,
+      "step": 1979
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.33012032009789e-05,
+      "loss": 0.8644,
+      "step": 1980
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.321377102957722e-05,
+      "loss": 0.2308,
+      "step": 1981
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.312634406952159e-05,
+      "loss": 0.8235,
+      "step": 1982
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.303892238794964e-05,
+      "loss": 0.8102,
+      "step": 1983
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.295150605199508e-05,
+      "loss": 0.2496,
+      "step": 1984
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.28640951287876e-05,
+      "loss": 0.7995,
+      "step": 1985
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.277668968545253e-05,
+      "loss": 0.8307,
+      "step": 1986
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.268928978911112e-05,
+      "loss": 0.7602,
+      "step": 1987
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.260189550688035e-05,
+      "loss": 0.778,
+      "step": 1988
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.251450690587288e-05,
+      "loss": 0.8302,
+      "step": 1989
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.2427124053197e-05,
+      "loss": 0.7953,
+      "step": 1990
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.233974701595655e-05,
+      "loss": 0.805,
+      "step": 1991
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.225237586125099e-05,
+      "loss": 0.7942,
+      "step": 1992
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.216501065617518e-05,
+      "loss": 0.808,
+      "step": 1993
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.207765146781946e-05,
+      "loss": 0.798,
+      "step": 1994
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.199029836326956e-05,
+      "loss": 0.8317,
+      "step": 1995
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.190295140960647e-05,
+      "loss": 0.8211,
+      "step": 1996
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.181561067390654e-05,
+      "loss": 0.8189,
+      "step": 1997
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.172827622324123e-05,
+      "loss": 0.2581,
+      "step": 1998
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.164094812467734e-05,
+      "loss": 0.805,
+      "step": 1999
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.155362644527668e-05,
+      "loss": 0.7886,
+      "step": 2000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.146631125209607e-05,
+      "loss": 0.8128,
+      "step": 2001
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.137900261218756e-05,
+      "loss": 0.7892,
+      "step": 2002
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.129170059259794e-05,
+      "loss": 0.8181,
+      "step": 2003
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.120440526036903e-05,
+      "loss": 0.8058,
+      "step": 2004
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.111711668253752e-05,
+      "loss": 0.2554,
+      "step": 2005
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.102983492613489e-05,
+      "loss": 0.7776,
+      "step": 2006
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.094256005818736e-05,
+      "loss": 0.8487,
+      "step": 2007
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.085529214571589e-05,
+      "loss": 0.8351,
+      "step": 2008
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.076803125573611e-05,
+      "loss": 0.2438,
+      "step": 2009
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.068077745525823e-05,
+      "loss": 0.8271,
+      "step": 2010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.0593530811287e-05,
+      "loss": 0.8252,
+      "step": 2011
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.050629139082175e-05,
+      "loss": 0.8665,
+      "step": 2012
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.041905926085618e-05,
+      "loss": 0.8375,
+      "step": 2013
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.033183448837844e-05,
+      "loss": 0.7913,
+      "step": 2014
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.024461714037094e-05,
+      "loss": 0.8438,
+      "step": 2015
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.015740728381054e-05,
+      "loss": 0.8382,
+      "step": 2016
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.007020498566829e-05,
+      "loss": 0.7723,
+      "step": 2017
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.998301031290931e-05,
+      "loss": 0.82,
+      "step": 2018
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.989582333249306e-05,
+      "loss": 0.843,
+      "step": 2019
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.980864411137302e-05,
+      "loss": 0.8036,
+      "step": 2020
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.97214727164966e-05,
+      "loss": 0.7993,
+      "step": 2021
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.963430921480533e-05,
+      "loss": 0.8441,
+      "step": 2022
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.954715367323468e-05,
+      "loss": 0.8308,
+      "step": 2023
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.946000615871392e-05,
+      "loss": 0.782,
+      "step": 2024
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.937286673816621e-05,
+      "loss": 0.7935,
+      "step": 2025
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.928573547850851e-05,
+      "loss": 0.8007,
+      "step": 2026
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.91986124466515e-05,
+      "loss": 0.8122,
+      "step": 2027
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.911149770949948e-05,
+      "loss": 0.8303,
+      "step": 2028
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.902439133395051e-05,
+      "loss": 0.8142,
+      "step": 2029
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.89372933868961e-05,
+      "loss": 0.2297,
+      "step": 2030
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.885020393522135e-05,
+      "loss": 0.818,
+      "step": 2031
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.876312304580482e-05,
+      "loss": 0.824,
+      "step": 2032
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.867605078551852e-05,
+      "loss": 0.8353,
+      "step": 2033
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.858898722122782e-05,
+      "loss": 0.7857,
+      "step": 2034
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.850193241979136e-05,
+      "loss": 0.8142,
+      "step": 2035
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.841488644806114e-05,
+      "loss": 0.7912,
+      "step": 2036
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.832784937288237e-05,
+      "loss": 0.7636,
+      "step": 2037
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.82408212610933e-05,
+      "loss": 0.275,
+      "step": 2038
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.815380217952542e-05,
+      "loss": 0.81,
+      "step": 2039
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.806679219500333e-05,
+      "loss": 0.7949,
+      "step": 2040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.79797913743445e-05,
+      "loss": 0.7741,
+      "step": 2041
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.789279978435945e-05,
+      "loss": 0.83,
+      "step": 2042
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.78058174918516e-05,
+      "loss": 0.8324,
+      "step": 2043
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.771884456361723e-05,
+      "loss": 0.8358,
+      "step": 2044
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.76318810664454e-05,
+      "loss": 0.8184,
+      "step": 2045
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.754492706711798e-05,
+      "loss": 0.8222,
+      "step": 2046
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.745798263240952e-05,
+      "loss": 0.8273,
+      "step": 2047
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.737104782908721e-05,
+      "loss": 0.8067,
+      "step": 2048
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.728412272391083e-05,
+      "loss": 0.7977,
+      "step": 2049
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.71972073836328e-05,
+      "loss": 0.7822,
+      "step": 2050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.711030187499795e-05,
+      "loss": 0.2457,
+      "step": 2051
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.702340626474355e-05,
+      "loss": 0.8047,
+      "step": 2052
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.69365206195994e-05,
+      "loss": 0.8197,
+      "step": 2053
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.68496450062875e-05,
+      "loss": 0.785,
+      "step": 2054
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.676277949152223e-05,
+      "loss": 0.8093,
+      "step": 2055
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.667592414201009e-05,
+      "loss": 0.8224,
+      "step": 2056
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.658907902445004e-05,
+      "loss": 0.7458,
+      "step": 2057
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.650224420553288e-05,
+      "loss": 0.8702,
+      "step": 2058
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.641541975194165e-05,
+      "loss": 0.8212,
+      "step": 2059
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.632860573035152e-05,
+      "loss": 0.844,
+      "step": 2060
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.624180220742946e-05,
+      "loss": 0.8111,
+      "step": 2061
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.615500924983442e-05,
+      "loss": 0.8424,
+      "step": 2062
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.60682269242174e-05,
+      "loss": 0.8804,
+      "step": 2063
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.598145529722104e-05,
+      "loss": 0.8187,
+      "step": 2064
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.589469443547984e-05,
+      "loss": 0.8566,
+      "step": 2065
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.580794440562002e-05,
+      "loss": 0.8112,
+      "step": 2066
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.572120527425955e-05,
+      "loss": 0.7883,
+      "step": 2067
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.563447710800794e-05,
+      "loss": 0.8298,
+      "step": 2068
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.554775997346629e-05,
+      "loss": 0.8195,
+      "step": 2069
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.546105393722731e-05,
+      "loss": 0.8,
+      "step": 2070
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.53743590658751e-05,
+      "loss": 0.8355,
+      "step": 2071
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.528767542598523e-05,
+      "loss": 0.8108,
+      "step": 2072
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.520100308412455e-05,
+      "loss": 0.8015,
+      "step": 2073
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.511434210685143e-05,
+      "loss": 0.8351,
+      "step": 2074
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.502769256071537e-05,
+      "loss": 0.8584,
+      "step": 2075
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.494105451225704e-05,
+      "loss": 0.8259,
+      "step": 2076
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.485442802800849e-05,
+      "loss": 0.8547,
+      "step": 2077
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.476781317449268e-05,
+      "loss": 0.7774,
+      "step": 2078
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.46812100182237e-05,
+      "loss": 0.8013,
+      "step": 2079
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.45946186257067e-05,
+      "loss": 0.2391,
+      "step": 2080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.45080390634378e-05,
+      "loss": 0.8289,
+      "step": 2081
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.442147139790397e-05,
+      "loss": 0.7688,
+      "step": 2082
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.433491569558309e-05,
+      "loss": 0.856,
+      "step": 2083
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.424837202294385e-05,
+      "loss": 0.8179,
+      "step": 2084
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.416184044644571e-05,
+      "loss": 0.7988,
+      "step": 2085
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.407532103253878e-05,
+      "loss": 0.7929,
+      "step": 2086
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.398881384766394e-05,
+      "loss": 0.8001,
+      "step": 2087
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.390231895825258e-05,
+      "loss": 0.8089,
+      "step": 2088
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.38158364307267e-05,
+      "loss": 0.7847,
+      "step": 2089
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.372936633149874e-05,
+      "loss": 0.8626,
+      "step": 2090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.364290872697173e-05,
+      "loss": 0.8304,
+      "step": 2091
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.3556463683539e-05,
+      "loss": 0.7999,
+      "step": 2092
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.347003126758417e-05,
+      "loss": 0.8025,
+      "step": 2093
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.338361154548137e-05,
+      "loss": 0.8131,
+      "step": 2094
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.329720458359481e-05,
+      "loss": 0.8004,
+      "step": 2095
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.321081044827895e-05,
+      "loss": 0.8268,
+      "step": 2096
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.312442920587833e-05,
+      "loss": 0.8229,
+      "step": 2097
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.303806092272781e-05,
+      "loss": 0.7863,
+      "step": 2098
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.295170566515204e-05,
+      "loss": 0.7818,
+      "step": 2099
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.286536349946581e-05,
+      "loss": 0.8652,
+      "step": 2100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.277903449197383e-05,
+      "loss": 0.7862,
+      "step": 2101
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.269271870897069e-05,
+      "loss": 0.2469,
+      "step": 2102
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.260641621674082e-05,
+      "loss": 0.8029,
+      "step": 2103
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.252012708155848e-05,
+      "loss": 0.7959,
+      "step": 2104
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.243385136968765e-05,
+      "loss": 0.8418,
+      "step": 2105
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.234758914738199e-05,
+      "loss": 0.7935,
+      "step": 2106
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.226134048088478e-05,
+      "loss": 0.8386,
+      "step": 2107
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.2175105436429e-05,
+      "loss": 0.809,
+      "step": 2108
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.208888408023704e-05,
+      "loss": 0.7965,
+      "step": 2109
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.200267647852082e-05,
+      "loss": 0.814,
+      "step": 2110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.191648269748174e-05,
+      "loss": 0.8856,
+      "step": 2111
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.183030280331054e-05,
+      "loss": 0.8195,
+      "step": 2112
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.174413686218732e-05,
+      "loss": 0.9066,
+      "step": 2113
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.16579849402814e-05,
+      "loss": 0.8063,
+      "step": 2114
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.157184710375149e-05,
+      "loss": 0.8164,
+      "step": 2115
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.14857234187453e-05,
+      "loss": 0.8093,
+      "step": 2116
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.139961395139974e-05,
+      "loss": 0.7518,
+      "step": 2117
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.131351876784094e-05,
+      "loss": 0.8125,
+      "step": 2118
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.122743793418383e-05,
+      "loss": 0.8575,
+      "step": 2119
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.114137151653242e-05,
+      "loss": 0.8337,
+      "step": 2120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.105531958097972e-05,
+      "loss": 0.8389,
+      "step": 2121
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.096928219360756e-05,
+      "loss": 0.8,
+      "step": 2122
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.088325942048653e-05,
+      "loss": 0.8269,
+      "step": 2123
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.07972513276761e-05,
+      "loss": 0.7595,
+      "step": 2124
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.071125798122444e-05,
+      "loss": 0.7817,
+      "step": 2125
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.062527944716838e-05,
+      "loss": 0.8133,
+      "step": 2126
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.053931579153333e-05,
+      "loss": 0.8219,
+      "step": 2127
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.045336708033341e-05,
+      "loss": 0.8615,
+      "step": 2128
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.036743337957112e-05,
+      "loss": 0.815,
+      "step": 2129
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.028151475523753e-05,
+      "loss": 0.7716,
+      "step": 2130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.019561127331201e-05,
+      "loss": 0.7984,
+      "step": 2131
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.010972299976248e-05,
+      "loss": 0.7631,
+      "step": 2132
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.002385000054506e-05,
+      "loss": 0.8154,
+      "step": 2133
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.99379923416041e-05,
+      "loss": 0.8308,
+      "step": 2134
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.985215008887237e-05,
+      "loss": 0.8353,
+      "step": 2135
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.976632330827056e-05,
+      "loss": 0.8101,
+      "step": 2136
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.968051206570758e-05,
+      "loss": 0.8341,
+      "step": 2137
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.959471642708055e-05,
+      "loss": 0.7976,
+      "step": 2138
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.950893645827437e-05,
+      "loss": 0.8496,
+      "step": 2139
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.942317222516204e-05,
+      "loss": 0.9023,
+      "step": 2140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.933742379360446e-05,
+      "loss": 0.8294,
+      "step": 2141
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.925169122945039e-05,
+      "loss": 0.2339,
+      "step": 2142
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.916597459853641e-05,
+      "loss": 0.8111,
+      "step": 2143
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.90802739666868e-05,
+      "loss": 0.8142,
+      "step": 2144
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.89945893997137e-05,
+      "loss": 0.8415,
+      "step": 2145
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.890892096341677e-05,
+      "loss": 0.8041,
+      "step": 2146
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.882326872358333e-05,
+      "loss": 0.8334,
+      "step": 2147
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.873763274598827e-05,
+      "loss": 0.8305,
+      "step": 2148
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.865201309639402e-05,
+      "loss": 0.7762,
+      "step": 2149
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.856640984055047e-05,
+      "loss": 0.8346,
+      "step": 2150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.848082304419478e-05,
+      "loss": 0.8315,
+      "step": 2151
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.839525277305166e-05,
+      "loss": 0.8124,
+      "step": 2152
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.83096990928331e-05,
+      "loss": 0.8283,
+      "step": 2153
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.822416206923816e-05,
+      "loss": 0.8077,
+      "step": 2154
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.81386417679534e-05,
+      "loss": 0.8007,
+      "step": 2155
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.805313825465231e-05,
+      "loss": 0.8389,
+      "step": 2156
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.796765159499558e-05,
+      "loss": 0.8224,
+      "step": 2157
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.788218185463093e-05,
+      "loss": 0.7823,
+      "step": 2158
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.779672909919313e-05,
+      "loss": 0.8174,
+      "step": 2159
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.771129339430389e-05,
+      "loss": 0.8525,
+      "step": 2160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.762587480557175e-05,
+      "loss": 0.7806,
+      "step": 2161
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.754047339859224e-05,
+      "loss": 0.8032,
+      "step": 2162
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.74550892389476e-05,
+      "loss": 0.7803,
+      "step": 2163
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.736972239220686e-05,
+      "loss": 0.8093,
+      "step": 2164
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.728437292392571e-05,
+      "loss": 0.8386,
+      "step": 2165
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.719904089964658e-05,
+      "loss": 0.8279,
+      "step": 2166
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.711372638489844e-05,
+      "loss": 0.7892,
+      "step": 2167
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.70284294451968e-05,
+      "loss": 0.252,
+      "step": 2168
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.694315014604376e-05,
+      "loss": 0.8229,
+      "step": 2169
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.685788855292781e-05,
+      "loss": 0.7933,
+      "step": 2170
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.677264473132379e-05,
+      "loss": 0.8331,
+      "step": 2171
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.668741874669297e-05,
+      "loss": 0.8031,
+      "step": 2172
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.660221066448297e-05,
+      "loss": 0.7966,
+      "step": 2173
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.651702055012752e-05,
+      "loss": 0.8254,
+      "step": 2174
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.643184846904662e-05,
+      "loss": 0.8106,
+      "step": 2175
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.634669448664654e-05,
+      "loss": 0.2683,
+      "step": 2176
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.626155866831944e-05,
+      "loss": 0.8175,
+      "step": 2177
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.617644107944366e-05,
+      "loss": 0.8193,
+      "step": 2178
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.609134178538352e-05,
+      "loss": 0.8228,
+      "step": 2179
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.60062608514893e-05,
+      "loss": 0.8141,
+      "step": 2180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.592119834309715e-05,
+      "loss": 0.7966,
+      "step": 2181
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.583615432552906e-05,
+      "loss": 0.8127,
+      "step": 2182
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.575112886409289e-05,
+      "loss": 0.8228,
+      "step": 2183
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.56661220240822e-05,
+      "loss": 0.7967,
+      "step": 2184
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.558113387077622e-05,
+      "loss": 0.7728,
+      "step": 2185
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.549616446943992e-05,
+      "loss": 0.8018,
+      "step": 2186
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.541121388532379e-05,
+      "loss": 0.8441,
+      "step": 2187
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.53262821836639e-05,
+      "loss": 0.8254,
+      "step": 2188
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.524136942968179e-05,
+      "loss": 0.2417,
+      "step": 2189
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.515647568858452e-05,
+      "loss": 0.7797,
+      "step": 2190
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.507160102556451e-05,
+      "loss": 0.8343,
+      "step": 2191
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.498674550579943e-05,
+      "loss": 0.808,
+      "step": 2192
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.490190919445248e-05,
+      "loss": 0.77,
+      "step": 2193
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.481709215667189e-05,
+      "loss": 0.2651,
+      "step": 2194
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.473229445759114e-05,
+      "loss": 0.7934,
+      "step": 2195
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.464751616232902e-05,
+      "loss": 0.8076,
+      "step": 2196
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.456275733598916e-05,
+      "loss": 0.7979,
+      "step": 2197
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.447801804366044e-05,
+      "loss": 0.8136,
+      "step": 2198
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.439329835041661e-05,
+      "loss": 0.8231,
+      "step": 2199
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.430859832131652e-05,
+      "loss": 0.8102,
+      "step": 2200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.422391802140377e-05,
+      "loss": 0.8254,
+      "step": 2201
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.413925751570684e-05,
+      "loss": 0.8542,
+      "step": 2202
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.40546168692391e-05,
+      "loss": 0.8114,
+      "step": 2203
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.396999614699857e-05,
+      "loss": 0.8554,
+      "step": 2204
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.388539541396802e-05,
+      "loss": 0.8099,
+      "step": 2205
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.380081473511484e-05,
+      "loss": 0.8157,
+      "step": 2206
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.371625417539106e-05,
+      "loss": 0.7726,
+      "step": 2207
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.363171379973322e-05,
+      "loss": 0.2868,
+      "step": 2208
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.354719367306233e-05,
+      "loss": 0.8497,
+      "step": 2209
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.346269386028398e-05,
+      "loss": 0.7997,
+      "step": 2210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.337821442628805e-05,
+      "loss": 0.7889,
+      "step": 2211
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.329375543594872e-05,
+      "loss": 0.8169,
+      "step": 2212
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.320931695412467e-05,
+      "loss": 0.7996,
+      "step": 2213
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.312489904565864e-05,
+      "loss": 0.7967,
+      "step": 2214
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.304050177537764e-05,
+      "loss": 0.8293,
+      "step": 2215
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.29561252080928e-05,
+      "loss": 0.7793,
+      "step": 2216
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.287176940859948e-05,
+      "loss": 0.2534,
+      "step": 2217
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.278743444167691e-05,
+      "loss": 0.7632,
+      "step": 2218
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.27031203720884e-05,
+      "loss": 0.8435,
+      "step": 2219
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.261882726458126e-05,
+      "loss": 0.7868,
+      "step": 2220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.253455518388668e-05,
+      "loss": 0.8348,
+      "step": 2221
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.245030419471964e-05,
+      "loss": 0.8074,
+      "step": 2222
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.236607436177893e-05,
+      "loss": 0.8152,
+      "step": 2223
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.228186574974724e-05,
+      "loss": 0.819,
+      "step": 2224
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.219767842329077e-05,
+      "loss": 0.8235,
+      "step": 2225
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.211351244705946e-05,
+      "loss": 0.8091,
+      "step": 2226
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.20293678856869e-05,
+      "loss": 0.8135,
+      "step": 2227
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.194524480379018e-05,
+      "loss": 0.8282,
+      "step": 2228
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.186114326596984e-05,
+      "loss": 0.8214,
+      "step": 2229
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.177706333681001e-05,
+      "loss": 0.784,
+      "step": 2230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.169300508087815e-05,
+      "loss": 0.8497,
+      "step": 2231
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.160896856272506e-05,
+      "loss": 0.226,
+      "step": 2232
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.152495384688482e-05,
+      "loss": 0.8324,
+      "step": 2233
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.14409609978749e-05,
+      "loss": 0.8236,
+      "step": 2234
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.135699008019585e-05,
+      "loss": 0.8551,
+      "step": 2235
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.127304115833142e-05,
+      "loss": 0.834,
+      "step": 2236
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.118911429674848e-05,
+      "loss": 0.8018,
+      "step": 2237
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.110520955989697e-05,
+      "loss": 0.7952,
+      "step": 2238
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.102132701220979e-05,
+      "loss": 0.808,
+      "step": 2239
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.093746671810282e-05,
+      "loss": 0.846,
+      "step": 2240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.08536287419749e-05,
+      "loss": 0.7917,
+      "step": 2241
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.076981314820766e-05,
+      "loss": 0.8278,
+      "step": 2242
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.068602000116558e-05,
+      "loss": 0.8031,
+      "step": 2243
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.060224936519592e-05,
+      "loss": 0.221,
+      "step": 2244
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.05185013046286e-05,
+      "loss": 0.7694,
+      "step": 2245
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.043477588377622e-05,
+      "loss": 0.8336,
+      "step": 2246
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.035107316693408e-05,
+      "loss": 0.7998,
+      "step": 2247
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.026739321837992e-05,
+      "loss": 0.8308,
+      "step": 2248
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.018373610237404e-05,
+      "loss": 0.8008,
+      "step": 2249
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.010010188315919e-05,
+      "loss": 0.7986,
+      "step": 2250
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.001649062496065e-05,
+      "loss": 0.2316,
+      "step": 2251
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.99329023919859e-05,
+      "loss": 0.7761,
+      "step": 2252
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.984933724842481e-05,
+      "loss": 0.8173,
+      "step": 2253
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.976579525844962e-05,
+      "loss": 0.8182,
+      "step": 2254
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.968227648621462e-05,
+      "loss": 0.7852,
+      "step": 2255
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.959878099585635e-05,
+      "loss": 0.819,
+      "step": 2256
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.951530885149344e-05,
+      "loss": 0.7894,
+      "step": 2257
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.943186011722669e-05,
+      "loss": 0.8267,
+      "step": 2258
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.934843485713882e-05,
+      "loss": 0.8155,
+      "step": 2259
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.926503313529454e-05,
+      "loss": 0.8258,
+      "step": 2260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.918165501574051e-05,
+      "loss": 0.8749,
+      "step": 2261
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.909830056250527e-05,
+      "loss": 0.8323,
+      "step": 2262
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.901496983959916e-05,
+      "loss": 0.8108,
+      "step": 2263
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.893166291101427e-05,
+      "loss": 0.7623,
+      "step": 2264
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.884837984072451e-05,
+      "loss": 0.7726,
+      "step": 2265
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.876512069268542e-05,
+      "loss": 0.8053,
+      "step": 2266
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.868188553083406e-05,
+      "loss": 0.8202,
+      "step": 2267
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.85986744190893e-05,
+      "loss": 0.7636,
+      "step": 2268
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.851548742135137e-05,
+      "loss": 0.7745,
+      "step": 2269
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.843232460150198e-05,
+      "loss": 0.8348,
+      "step": 2270
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.834918602340438e-05,
+      "loss": 0.8479,
+      "step": 2271
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.826607175090314e-05,
+      "loss": 0.8301,
+      "step": 2272
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.818298184782412e-05,
+      "loss": 0.8045,
+      "step": 2273
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.809991637797455e-05,
+      "loss": 0.7923,
+      "step": 2274
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.801687540514286e-05,
+      "loss": 0.7784,
+      "step": 2275
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.793385899309866e-05,
+      "loss": 0.8389,
+      "step": 2276
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.78508672055927e-05,
+      "loss": 0.7828,
+      "step": 2277
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.776790010635686e-05,
+      "loss": 0.8108,
+      "step": 2278
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.768495775910401e-05,
+      "loss": 0.8115,
+      "step": 2279
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.760204022752805e-05,
+      "loss": 0.8013,
+      "step": 2280
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.751914757530376e-05,
+      "loss": 0.8447,
+      "step": 2281
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.74362798660869e-05,
+      "loss": 0.8079,
+      "step": 2282
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.735343716351405e-05,
+      "loss": 0.8577,
+      "step": 2283
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.727061953120249e-05,
+      "loss": 0.8088,
+      "step": 2284
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.718782703275042e-05,
+      "loss": 0.7917,
+      "step": 2285
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.710505973173664e-05,
+      "loss": 0.8725,
+      "step": 2286
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.702231769172052e-05,
+      "loss": 0.7777,
+      "step": 2287
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.693960097624223e-05,
+      "loss": 0.8488,
+      "step": 2288
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.685690964882236e-05,
+      "loss": 0.8169,
+      "step": 2289
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.677424377296196e-05,
+      "loss": 0.8062,
+      "step": 2290
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.669160341214264e-05,
+      "loss": 0.8162,
+      "step": 2291
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.660898862982641e-05,
+      "loss": 0.8271,
+      "step": 2292
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.652639948945556e-05,
+      "loss": 0.8031,
+      "step": 2293
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.644383605445275e-05,
+      "loss": 0.8078,
+      "step": 2294
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.63612983882209e-05,
+      "loss": 0.7876,
+      "step": 2295
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.62787865541431e-05,
+      "loss": 0.2528,
+      "step": 2296
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.619630061558265e-05,
+      "loss": 0.7963,
+      "step": 2297
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.611384063588291e-05,
+      "loss": 0.7885,
+      "step": 2298
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.603140667836738e-05,
+      "loss": 0.7994,
+      "step": 2299
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.594899880633952e-05,
+      "loss": 0.8016,
+      "step": 2300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.586661708308272e-05,
+      "loss": 0.8132,
+      "step": 2301
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.578426157186044e-05,
+      "loss": 0.7872,
+      "step": 2302
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.570193233591584e-05,
+      "loss": 0.7948,
+      "step": 2303
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5619629438472e-05,
+      "loss": 0.8026,
+      "step": 2304
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.553735294273176e-05,
+      "loss": 0.8284,
+      "step": 2305
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.545510291187766e-05,
+      "loss": 0.8129,
+      "step": 2306
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.537287940907194e-05,
+      "loss": 0.8315,
+      "step": 2307
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.52906824974564e-05,
+      "loss": 0.7928,
+      "step": 2308
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.520851224015261e-05,
+      "loss": 0.7586,
+      "step": 2309
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.512636870026143e-05,
+      "loss": 0.779,
+      "step": 2310
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.504425194086335e-05,
+      "loss": 0.7879,
+      "step": 2311
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.496216202501826e-05,
+      "loss": 0.8162,
+      "step": 2312
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.488009901576544e-05,
+      "loss": 0.8372,
+      "step": 2313
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.47980629761235e-05,
+      "loss": 0.8275,
+      "step": 2314
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.47160539690903e-05,
+      "loss": 0.8239,
+      "step": 2315
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.463407205764305e-05,
+      "loss": 0.8284,
+      "step": 2316
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.455211730473807e-05,
+      "loss": 0.7545,
+      "step": 2317
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.447018977331078e-05,
+      "loss": 0.826,
+      "step": 2318
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.438828952627582e-05,
+      "loss": 0.8187,
+      "step": 2319
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.430641662652681e-05,
+      "loss": 0.835,
+      "step": 2320
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.422457113693633e-05,
+      "loss": 0.7871,
+      "step": 2321
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.414275312035598e-05,
+      "loss": 0.7762,
+      "step": 2322
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.406096263961625e-05,
+      "loss": 0.816,
+      "step": 2323
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.397919975752647e-05,
+      "loss": 0.8185,
+      "step": 2324
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.389746453687468e-05,
+      "loss": 0.2179,
+      "step": 2325
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.381575704042792e-05,
+      "loss": 0.8316,
+      "step": 2326
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.37340773309317e-05,
+      "loss": 0.837,
+      "step": 2327
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.365242547111027e-05,
+      "loss": 0.8552,
+      "step": 2328
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.357080152366661e-05,
+      "loss": 0.2397,
+      "step": 2329
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.348920555128208e-05,
+      "loss": 0.8176,
+      "step": 2330
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.340763761661665e-05,
+      "loss": 0.8091,
+      "step": 2331
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.332609778230876e-05,
+      "loss": 0.8012,
+      "step": 2332
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.324458611097528e-05,
+      "loss": 0.84,
+      "step": 2333
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.316310266521142e-05,
+      "loss": 0.8292,
+      "step": 2334
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.308164750759069e-05,
+      "loss": 0.7889,
+      "step": 2335
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.300022070066498e-05,
+      "loss": 0.8318,
+      "step": 2336
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.291882230696432e-05,
+      "loss": 0.8472,
+      "step": 2337
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.283745238899692e-05,
+      "loss": 0.8116,
+      "step": 2338
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.275611100924916e-05,
+      "loss": 0.8197,
+      "step": 2339
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.267479823018553e-05,
+      "loss": 0.8158,
+      "step": 2340
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.259351411424849e-05,
+      "loss": 0.8462,
+      "step": 2341
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.251225872385845e-05,
+      "loss": 0.7883,
+      "step": 2342
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.243103212141394e-05,
+      "loss": 0.7837,
+      "step": 2343
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.234983436929124e-05,
+      "loss": 0.8053,
+      "step": 2344
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.226866552984443e-05,
+      "loss": 0.8087,
+      "step": 2345
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.218752566540554e-05,
+      "loss": 0.8344,
+      "step": 2346
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.21064148382843e-05,
+      "loss": 0.8304,
+      "step": 2347
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.202533311076803e-05,
+      "loss": 0.8153,
+      "step": 2348
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.19442805451218e-05,
+      "loss": 0.7841,
+      "step": 2349
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.186325720358833e-05,
+      "loss": 0.8513,
+      "step": 2350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.17822631483878e-05,
+      "loss": 0.8533,
+      "step": 2351
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.170129844171792e-05,
+      "loss": 0.7824,
+      "step": 2352
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.162036314575394e-05,
+      "loss": 0.8033,
+      "step": 2353
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.153945732264845e-05,
+      "loss": 0.7696,
+      "step": 2354
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.145858103453141e-05,
+      "loss": 0.847,
+      "step": 2355
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.137773434351009e-05,
+      "loss": 0.7718,
+      "step": 2356
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.12969173116691e-05,
+      "loss": 0.8276,
+      "step": 2357
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.121613000107023e-05,
+      "loss": 0.8049,
+      "step": 2358
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.113537247375238e-05,
+      "loss": 0.7661,
+      "step": 2359
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.105464479173174e-05,
+      "loss": 0.8175,
+      "step": 2360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.097394701700145e-05,
+      "loss": 0.7703,
+      "step": 2361
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.089327921153165e-05,
+      "loss": 0.7635,
+      "step": 2362
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0812641437269626e-05,
+      "loss": 0.7771,
+      "step": 2363
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.073203375613949e-05,
+      "loss": 0.8803,
+      "step": 2364
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0651456230042236e-05,
+      "loss": 0.8069,
+      "step": 2365
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0570908920855706e-05,
+      "loss": 0.8325,
+      "step": 2366
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.049039189043466e-05,
+      "loss": 0.8195,
+      "step": 2367
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.040990520061043e-05,
+      "loss": 0.7908,
+      "step": 2368
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.03294489131911e-05,
+      "loss": 0.8034,
+      "step": 2369
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.024902308996152e-05,
+      "loss": 0.7276,
+      "step": 2370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.016862779268301e-05,
+      "loss": 0.8224,
+      "step": 2371
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0088263083093496e-05,
+      "loss": 0.7775,
+      "step": 2372
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.000792902290741e-05,
+      "loss": 0.81,
+      "step": 2373
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9927625673815665e-05,
+      "loss": 0.7721,
+      "step": 2374
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.98473530974856e-05,
+      "loss": 0.8021,
+      "step": 2375
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.976711135556086e-05,
+      "loss": 0.7857,
+      "step": 2376
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.968690050966148e-05,
+      "loss": 0.7884,
+      "step": 2377
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.960672062138374e-05,
+      "loss": 0.8292,
+      "step": 2378
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.952657175230012e-05,
+      "loss": 0.2498,
+      "step": 2379
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.944645396395934e-05,
+      "loss": 0.812,
+      "step": 2380
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9366367317886204e-05,
+      "loss": 0.7655,
+      "step": 2381
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9286311875581656e-05,
+      "loss": 0.79,
+      "step": 2382
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9206287698522546e-05,
+      "loss": 0.7954,
+      "step": 2383
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9126294848161914e-05,
+      "loss": 0.844,
+      "step": 2384
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.904633338592857e-05,
+      "loss": 0.8192,
+      "step": 2385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.896640337322725e-05,
+      "loss": 0.8339,
+      "step": 2386
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8886504871438694e-05,
+      "loss": 0.8146,
+      "step": 2387
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.880663794191925e-05,
+      "loss": 0.8164,
+      "step": 2388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.87268026460011e-05,
+      "loss": 0.8534,
+      "step": 2389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.864699904499211e-05,
+      "loss": 0.8415,
+      "step": 2390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8567227200175865e-05,
+      "loss": 0.7926,
+      "step": 2391
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8487487172811516e-05,
+      "loss": 0.8054,
+      "step": 2392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.840777902413376e-05,
+      "loss": 0.7916,
+      "step": 2393
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.832810281535288e-05,
+      "loss": 0.7987,
+      "step": 2394
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.82484586076546e-05,
+      "loss": 0.8203,
+      "step": 2395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8168846462199964e-05,
+      "loss": 0.7739,
+      "step": 2396
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8089266440125645e-05,
+      "loss": 0.8471,
+      "step": 2397
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.800971860254343e-05,
+      "loss": 0.8076,
+      "step": 2398
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.793020301054041e-05,
+      "loss": 0.8405,
+      "step": 2399
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7850719725178994e-05,
+      "loss": 0.8333,
+      "step": 2400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.777126880749679e-05,
+      "loss": 0.8103,
+      "step": 2401
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.769185031850644e-05,
+      "loss": 0.7906,
+      "step": 2402
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.761246431919576e-05,
+      "loss": 0.7914,
+      "step": 2403
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.753311087052768e-05,
+      "loss": 0.8267,
+      "step": 2404
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.745379003343998e-05,
+      "loss": 0.8016,
+      "step": 2405
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7374501868845544e-05,
+      "loss": 0.2504,
+      "step": 2406
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.729524643763202e-05,
+      "loss": 0.7907,
+      "step": 2407
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7216023800662046e-05,
+      "loss": 0.8283,
+      "step": 2408
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7136834018773047e-05,
+      "loss": 0.8332,
+      "step": 2409
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.705767715277714e-05,
+      "loss": 0.795,
+      "step": 2410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6978553263461265e-05,
+      "loss": 0.8183,
+      "step": 2411
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6899462411587027e-05,
+      "loss": 0.8253,
+      "step": 2412
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.682040465789056e-05,
+      "loss": 0.7968,
+      "step": 2413
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6741380063082715e-05,
+      "loss": 0.8499,
+      "step": 2414
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.666238868784875e-05,
+      "loss": 0.7796,
+      "step": 2415
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6583430592848565e-05,
+      "loss": 0.8333,
+      "step": 2416
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6504505838716325e-05,
+      "loss": 0.7612,
+      "step": 2417
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.642561448606072e-05,
+      "loss": 0.2541,
+      "step": 2418
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.634675659546481e-05,
+      "loss": 0.7884,
+      "step": 2419
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.626793222748581e-05,
+      "loss": 0.7955,
+      "step": 2420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6189141442655325e-05,
+      "loss": 0.8073,
+      "step": 2421
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6110384301479165e-05,
+      "loss": 0.8075,
+      "step": 2422
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6031660864437196e-05,
+      "loss": 0.803,
+      "step": 2423
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.595297119198352e-05,
+      "loss": 0.7937,
+      "step": 2424
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.587431534454628e-05,
+      "loss": 0.2553,
+      "step": 2425
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.579569338252758e-05,
+      "loss": 0.8376,
+      "step": 2426
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.571710536630357e-05,
+      "loss": 0.8149,
+      "step": 2427
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.563855135622435e-05,
+      "loss": 0.7397,
+      "step": 2428
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5560031412613846e-05,
+      "loss": 0.8332,
+      "step": 2429
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.548154559576979e-05,
+      "loss": 0.7748,
+      "step": 2430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5403093965963806e-05,
+      "loss": 0.8187,
+      "step": 2431
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.532467658344127e-05,
+      "loss": 0.7953,
+      "step": 2432
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.524629350842111e-05,
+      "loss": 0.7525,
+      "step": 2433
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.516794480109606e-05,
+      "loss": 0.7808,
+      "step": 2434
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5089630521632465e-05,
+      "loss": 0.8083,
+      "step": 2435
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.501135073017008e-05,
+      "loss": 0.8164,
+      "step": 2436
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.493310548682229e-05,
+      "loss": 0.771,
+      "step": 2437
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.485489485167602e-05,
+      "loss": 0.8162,
+      "step": 2438
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.477671888479142e-05,
+      "loss": 0.8081,
+      "step": 2439
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4698577646202224e-05,
+      "loss": 0.8437,
+      "step": 2440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.46204711959153e-05,
+      "loss": 0.803,
+      "step": 2441
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.454239959391095e-05,
+      "loss": 0.7635,
+      "step": 2442
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4464362900142715e-05,
+      "loss": 0.8312,
+      "step": 2443
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.43863611745372e-05,
+      "loss": 0.8315,
+      "step": 2444
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.430839447699427e-05,
+      "loss": 0.7911,
+      "step": 2445
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.42304628673869e-05,
+      "loss": 0.7995,
+      "step": 2446
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4152566405561036e-05,
+      "loss": 0.8165,
+      "step": 2447
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4074705151335615e-05,
+      "loss": 0.8272,
+      "step": 2448
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.399687916450272e-05,
+      "loss": 0.2217,
+      "step": 2449
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.391908850482719e-05,
+      "loss": 0.7766,
+      "step": 2450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3841333232046654e-05,
+      "loss": 0.798,
+      "step": 2451
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.376361340587188e-05,
+      "loss": 0.7745,
+      "step": 2452
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.368592908598611e-05,
+      "loss": 0.7851,
+      "step": 2453
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.360828033204539e-05,
+      "loss": 0.7967,
+      "step": 2454
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.353066720367862e-05,
+      "loss": 0.7875,
+      "step": 2455
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3453089760487154e-05,
+      "loss": 0.8034,
+      "step": 2456
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.337554806204497e-05,
+      "loss": 0.8149,
+      "step": 2457
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.329804216789865e-05,
+      "loss": 0.8,
+      "step": 2458
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3220572137567326e-05,
+      "loss": 0.2616,
+      "step": 2459
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3143138030542436e-05,
+      "loss": 0.7913,
+      "step": 2460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.306573990628796e-05,
+      "loss": 0.7953,
+      "step": 2461
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2988377824240266e-05,
+      "loss": 0.7618,
+      "step": 2462
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.291105184380787e-05,
+      "loss": 0.8216,
+      "step": 2463
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.283376202437178e-05,
+      "loss": 0.8178,
+      "step": 2464
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.275650842528507e-05,
+      "loss": 0.8247,
+      "step": 2465
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.267929110587307e-05,
+      "loss": 0.8277,
+      "step": 2466
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.260211012543331e-05,
+      "loss": 0.83,
+      "step": 2467
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2524965543235264e-05,
+      "loss": 0.7992,
+      "step": 2468
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2447857418520564e-05,
+      "loss": 0.8267,
+      "step": 2469
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.237078581050288e-05,
+      "loss": 0.8248,
+      "step": 2470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.229375077836769e-05,
+      "loss": 0.7903,
+      "step": 2471
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.221675238127255e-05,
+      "loss": 0.777,
+      "step": 2472
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.213979067834673e-05,
+      "loss": 0.2568,
+      "step": 2473
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.206286572869148e-05,
+      "loss": 0.795,
+      "step": 2474
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1985977591379664e-05,
+      "loss": 0.7757,
+      "step": 2475
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.190912632545599e-05,
+      "loss": 0.8112,
+      "step": 2476
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.183231198993688e-05,
+      "loss": 0.7827,
+      "step": 2477
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.175553464381023e-05,
+      "loss": 0.848,
+      "step": 2478
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.167879434603569e-05,
+      "loss": 0.8337,
+      "step": 2479
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.160209115554444e-05,
+      "loss": 0.8486,
+      "step": 2480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1525425131239056e-05,
+      "loss": 0.7768,
+      "step": 2481
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1448796331993666e-05,
+      "loss": 0.8414,
+      "step": 2482
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1372204816653846e-05,
+      "loss": 0.8055,
+      "step": 2483
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1295650644036406e-05,
+      "loss": 0.2561,
+      "step": 2484
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1219133872929607e-05,
+      "loss": 0.8077,
+      "step": 2485
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1142654562092976e-05,
+      "loss": 0.7863,
+      "step": 2486
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1066212770257185e-05,
+      "loss": 0.818,
+      "step": 2487
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0989808556124075e-05,
+      "loss": 0.8122,
+      "step": 2488
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0913441978366874e-05,
+      "loss": 0.8365,
+      "step": 2489
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.083711309562963e-05,
+      "loss": 0.2558,
+      "step": 2490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.076082196652754e-05,
+      "loss": 0.8103,
+      "step": 2491
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0684568649646845e-05,
+      "loss": 0.2389,
+      "step": 2492
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0608353203544754e-05,
+      "loss": 0.8156,
+      "step": 2493
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0532175686749305e-05,
+      "loss": 0.2257,
+      "step": 2494
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.045603615775949e-05,
+      "loss": 0.7917,
+      "step": 2495
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0379934675045145e-05,
+      "loss": 0.8318,
+      "step": 2496
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.030387129704679e-05,
+      "loss": 0.8167,
+      "step": 2497
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.022784608217579e-05,
+      "loss": 0.8112,
+      "step": 2498
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.015185908881409e-05,
+      "loss": 0.7762,
+      "step": 2499
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0075910375314404e-05,
+      "loss": 0.8065,
+      "step": 2500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.000000000000002e-05,
+      "loss": 0.8252,
+      "step": 2501
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.992412802116469e-05,
+      "loss": 0.8119,
+      "step": 2502
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9848294497072786e-05,
+      "loss": 0.7777,
+      "step": 2503
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9772499485959155e-05,
+      "loss": 0.7796,
+      "step": 2504
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.969674304602895e-05,
+      "loss": 0.7904,
+      "step": 2505
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.962102523545782e-05,
+      "loss": 0.807,
+      "step": 2506
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.954534611239175e-05,
+      "loss": 0.7717,
+      "step": 2507
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9469705734946946e-05,
+      "loss": 0.7972,
+      "step": 2508
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.939410416120981e-05,
+      "loss": 0.799,
+      "step": 2509
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.931854144923717e-05,
+      "loss": 0.83,
+      "step": 2510
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.92430176570558e-05,
+      "loss": 0.8424,
+      "step": 2511
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9167532842662576e-05,
+      "loss": 0.8517,
+      "step": 2512
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.909208706402468e-05,
+      "loss": 0.8095,
+      "step": 2513
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9016680379079085e-05,
+      "loss": 0.8105,
+      "step": 2514
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8941312845732756e-05,
+      "loss": 0.8142,
+      "step": 2515
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8865984521862697e-05,
+      "loss": 0.8187,
+      "step": 2516
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.879069546531581e-05,
+      "loss": 0.8054,
+      "step": 2517
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8715445733908696e-05,
+      "loss": 0.8526,
+      "step": 2518
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8640235385427894e-05,
+      "loss": 0.7515,
+      "step": 2519
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8565064477629683e-05,
+      "loss": 0.7567,
+      "step": 2520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.848993306823997e-05,
+      "loss": 0.7976,
+      "step": 2521
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8414841214954446e-05,
+      "loss": 0.2636,
+      "step": 2522
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.833978897543829e-05,
+      "loss": 0.8262,
+      "step": 2523
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8264776407326394e-05,
+      "loss": 0.779,
+      "step": 2524
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8189803568223154e-05,
+      "loss": 0.849,
+      "step": 2525
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.811487051570235e-05,
+      "loss": 0.805,
+      "step": 2526
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.803997730730737e-05,
+      "loss": 0.7983,
+      "step": 2527
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7965124000550854e-05,
+      "loss": 0.7716,
+      "step": 2528
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.78903106529149e-05,
+      "loss": 0.2574,
+      "step": 2529
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7815537321850935e-05,
+      "loss": 0.2565,
+      "step": 2530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7740804064779534e-05,
+      "loss": 0.8164,
+      "step": 2531
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7666110939090644e-05,
+      "loss": 0.7798,
+      "step": 2532
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.759145800214326e-05,
+      "loss": 0.2416,
+      "step": 2533
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.751684531126562e-05,
+      "loss": 0.7936,
+      "step": 2534
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.744227292375505e-05,
+      "loss": 0.8161,
+      "step": 2535
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.736774089687781e-05,
+      "loss": 0.8659,
+      "step": 2536
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.729324928786929e-05,
+      "loss": 0.8134,
+      "step": 2537
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.721879815393385e-05,
+      "loss": 0.8064,
+      "step": 2538
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.714438755224463e-05,
+      "loss": 0.7725,
+      "step": 2539
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.707001753994377e-05,
+      "loss": 0.8254,
+      "step": 2540
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.699568817414224e-05,
+      "loss": 0.8467,
+      "step": 2541
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.692139951191973e-05,
+      "loss": 0.2219,
+      "step": 2542
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6847151610324623e-05,
+      "loss": 0.2363,
+      "step": 2543
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.677294452637422e-05,
+      "loss": 0.2356,
+      "step": 2544
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.669877831705427e-05,
+      "loss": 0.7844,
+      "step": 2545
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.662465303931912e-05,
+      "loss": 0.8204,
+      "step": 2546
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.655056875009192e-05,
+      "loss": 0.8112,
+      "step": 2547
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.64765255062641e-05,
+      "loss": 0.8752,
+      "step": 2548
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6402523364695624e-05,
+      "loss": 0.7799,
+      "step": 2549
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.632856238221498e-05,
+      "loss": 0.7819,
+      "step": 2550
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.625464261561902e-05,
+      "loss": 0.8784,
+      "step": 2551
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6180764121672825e-05,
+      "loss": 0.8199,
+      "step": 2552
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.610692695710994e-05,
+      "loss": 0.8106,
+      "step": 2553
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6033131178632136e-05,
+      "loss": 0.7993,
+      "step": 2554
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5959376842909306e-05,
+      "loss": 0.8431,
+      "step": 2555
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5885664006579645e-05,
+      "loss": 0.7822,
+      "step": 2556
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.581199272624936e-05,
+      "loss": 0.7966,
+      "step": 2557
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.573836305849284e-05,
+      "loss": 0.8121,
+      "step": 2558
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5664775059852515e-05,
+      "loss": 0.8033,
+      "step": 2559
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.559122878683872e-05,
+      "loss": 0.2336,
+      "step": 2560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5517724295929864e-05,
+      "loss": 0.8191,
+      "step": 2561
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5444261643572226e-05,
+      "loss": 0.7373,
+      "step": 2562
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.53708408861799e-05,
+      "loss": 0.7868,
+      "step": 2563
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.529746208013489e-05,
+      "loss": 0.7602,
+      "step": 2564
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.522412528178698e-05,
+      "loss": 0.7849,
+      "step": 2565
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5150830547453627e-05,
+      "loss": 0.7847,
+      "step": 2566
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.507757793341997e-05,
+      "loss": 0.8184,
+      "step": 2567
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.500436749593899e-05,
+      "loss": 0.755,
+      "step": 2568
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.493119929123109e-05,
+      "loss": 0.7732,
+      "step": 2569
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.485807337548425e-05,
+      "loss": 0.7739,
+      "step": 2570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.478498980485405e-05,
+      "loss": 0.785,
+      "step": 2571
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4711948635463584e-05,
+      "loss": 0.7933,
+      "step": 2572
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.463894992340325e-05,
+      "loss": 0.8019,
+      "step": 2573
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.456599372473094e-05,
+      "loss": 0.7654,
+      "step": 2574
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.449308009547193e-05,
+      "loss": 0.8084,
+      "step": 2575
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.442020909161867e-05,
+      "loss": 0.8028,
+      "step": 2576
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4347380769130995e-05,
+      "loss": 0.8138,
+      "step": 2577
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4274595183935974e-05,
+      "loss": 0.827,
+      "step": 2578
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4201852391927725e-05,
+      "loss": 0.7909,
+      "step": 2579
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.412915244896764e-05,
+      "loss": 0.7885,
+      "step": 2580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.405649541088419e-05,
+      "loss": 0.7644,
+      "step": 2581
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3983881333472784e-05,
+      "loss": 0.8023,
+      "step": 2582
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3911310272495996e-05,
+      "loss": 0.8382,
+      "step": 2583
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.383878228368321e-05,
+      "loss": 0.8032,
+      "step": 2584
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.376629742273091e-05,
+      "loss": 0.8465,
+      "step": 2585
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.369385574530227e-05,
+      "loss": 0.7662,
+      "step": 2586
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.362145730702744e-05,
+      "loss": 0.8135,
+      "step": 2587
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.354910216350335e-05,
+      "loss": 0.7506,
+      "step": 2588
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.347679037029361e-05,
+      "loss": 0.7533,
+      "step": 2589
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.340452198292863e-05,
+      "loss": 0.2493,
+      "step": 2590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.333229705690538e-05,
+      "loss": 0.7948,
+      "step": 2591
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.326011564768755e-05,
+      "loss": 0.7921,
+      "step": 2592
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.318797781070543e-05,
+      "loss": 0.8136,
+      "step": 2593
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.31158836013557e-05,
+      "loss": 0.8084,
+      "step": 2594
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.304383307500169e-05,
+      "loss": 0.7981,
+      "step": 2595
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.297182628697317e-05,
+      "loss": 0.8309,
+      "step": 2596
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.28998632925662e-05,
+      "loss": 0.8021,
+      "step": 2597
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.282794414704333e-05,
+      "loss": 0.7834,
+      "step": 2598
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.275606890563345e-05,
+      "loss": 0.7324,
+      "step": 2599
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.268423762353163e-05,
+      "loss": 0.8197,
+      "step": 2600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.261245035589917e-05,
+      "loss": 0.8186,
+      "step": 2601
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.254070715786378e-05,
+      "loss": 0.7966,
+      "step": 2602
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2469008084519114e-05,
+      "loss": 0.7823,
+      "step": 2603
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2397353190924926e-05,
+      "loss": 0.815,
+      "step": 2604
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2325742532107306e-05,
+      "loss": 0.814,
+      "step": 2605
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.225417616305809e-05,
+      "loss": 0.7725,
+      "step": 2606
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.218265413873519e-05,
+      "loss": 0.776,
+      "step": 2607
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2111176514062526e-05,
+      "loss": 0.8018,
+      "step": 2608
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2039743343929896e-05,
+      "loss": 0.804,
+      "step": 2609
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.196835468319289e-05,
+      "loss": 0.794,
+      "step": 2610
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.189701058667301e-05,
+      "loss": 0.749,
+      "step": 2611
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.182571110915753e-05,
+      "loss": 0.7766,
+      "step": 2612
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1754456305399336e-05,
+      "loss": 0.8014,
+      "step": 2613
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.16832462301172e-05,
+      "loss": 0.8266,
+      "step": 2614
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.161208093799537e-05,
+      "loss": 0.7946,
+      "step": 2615
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.15409604836838e-05,
+      "loss": 0.8439,
+      "step": 2616
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.146988492179803e-05,
+      "loss": 0.8204,
+      "step": 2617
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1398854306919036e-05,
+      "loss": 0.8204,
+      "step": 2618
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.132786869359333e-05,
+      "loss": 0.8114,
+      "step": 2619
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.125692813633292e-05,
+      "loss": 0.7752,
+      "step": 2620
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.118603268961508e-05,
+      "loss": 0.7817,
+      "step": 2621
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.111518240788258e-05,
+      "loss": 0.8078,
+      "step": 2622
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.104437734554345e-05,
+      "loss": 0.7953,
+      "step": 2623
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.097361755697097e-05,
+      "loss": 0.8102,
+      "step": 2624
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.090290309650361e-05,
+      "loss": 0.799,
+      "step": 2625
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0832234018445225e-05,
+      "loss": 0.8245,
+      "step": 2626
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.076161037706463e-05,
+      "loss": 0.7792,
+      "step": 2627
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.069103222659576e-05,
+      "loss": 0.7824,
+      "step": 2628
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.062049962123772e-05,
+      "loss": 0.7702,
+      "step": 2629
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.055001261515461e-05,
+      "loss": 0.8531,
+      "step": 2630
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.047957126247541e-05,
+      "loss": 0.8333,
+      "step": 2631
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.040917561729416e-05,
+      "loss": 0.8471,
+      "step": 2632
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0338825733669794e-05,
+      "loss": 0.2502,
+      "step": 2633
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0268521665626e-05,
+      "loss": 0.8084,
+      "step": 2634
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.019826346715138e-05,
+      "loss": 0.7216,
+      "step": 2635
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.012805119219932e-05,
+      "loss": 0.7931,
+      "step": 2636
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.005788489468784e-05,
+      "loss": 0.2359,
+      "step": 2637
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.998776462849973e-05,
+      "loss": 0.7806,
+      "step": 2638
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.991769044748249e-05,
+      "loss": 0.8408,
+      "step": 2639
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.984766240544805e-05,
+      "loss": 0.8086,
+      "step": 2640
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9777680556173104e-05,
+      "loss": 0.8159,
+      "step": 2641
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.970774495339871e-05,
+      "loss": 0.8264,
+      "step": 2642
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9637855650830544e-05,
+      "loss": 0.7958,
+      "step": 2643
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.956801270213862e-05,
+      "loss": 0.2399,
+      "step": 2644
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.949821616095743e-05,
+      "loss": 0.7554,
+      "step": 2645
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.942846608088583e-05,
+      "loss": 0.7994,
+      "step": 2646
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.93587625154869e-05,
+      "loss": 0.7728,
+      "step": 2647
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.928910551828816e-05,
+      "loss": 0.7893,
+      "step": 2648
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.921949514278119e-05,
+      "loss": 0.803,
+      "step": 2649
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9149931442421904e-05,
+      "loss": 0.8158,
+      "step": 2650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.908041447063034e-05,
+      "loss": 0.78,
+      "step": 2651
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.90109442807906e-05,
+      "loss": 0.8099,
+      "step": 2652
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.89415209262509e-05,
+      "loss": 0.8371,
+      "step": 2653
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8872144460323535e-05,
+      "loss": 0.7726,
+      "step": 2654
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.880281493628467e-05,
+      "loss": 0.8084,
+      "step": 2655
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.873353240737455e-05,
+      "loss": 0.752,
+      "step": 2656
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.866429692679729e-05,
+      "loss": 0.7924,
+      "step": 2657
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.859510854772086e-05,
+      "loss": 0.7824,
+      "step": 2658
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8525967323276955e-05,
+      "loss": 0.7958,
+      "step": 2659
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8456873306561336e-05,
+      "loss": 0.7865,
+      "step": 2660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.838782655063325e-05,
+      "loss": 0.7917,
+      "step": 2661
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.83188271085157e-05,
+      "loss": 0.8062,
+      "step": 2662
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.824987503319551e-05,
+      "loss": 0.7982,
+      "step": 2663
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.818097037762297e-05,
+      "loss": 0.7917,
+      "step": 2664
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.811211319471196e-05,
+      "loss": 0.8001,
+      "step": 2665
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.804330353733998e-05,
+      "loss": 0.8523,
+      "step": 2666
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7974541458348046e-05,
+      "loss": 0.7697,
+      "step": 2667
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7905827010540504e-05,
+      "loss": 0.8275,
+      "step": 2668
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7837160246685276e-05,
+      "loss": 0.8207,
+      "step": 2669
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.776854121951361e-05,
+      "loss": 0.7522,
+      "step": 2670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.769996998172002e-05,
+      "loss": 0.8369,
+      "step": 2671
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7631446585962424e-05,
+      "loss": 0.8142,
+      "step": 2672
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.756297108486202e-05,
+      "loss": 0.8284,
+      "step": 2673
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.749454353100307e-05,
+      "loss": 0.8045,
+      "step": 2674
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.742616397693322e-05,
+      "loss": 0.7715,
+      "step": 2675
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7357832475163045e-05,
+      "loss": 0.7976,
+      "step": 2676
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7289549078166386e-05,
+      "loss": 0.7834,
+      "step": 2677
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.722131383838011e-05,
+      "loss": 0.8052,
+      "step": 2678
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.715312680820402e-05,
+      "loss": 0.82,
+      "step": 2679
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.708498804000097e-05,
+      "loss": 0.2384,
+      "step": 2680
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.701689758609678e-05,
+      "loss": 0.796,
+      "step": 2681
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6948855498780086e-05,
+      "loss": 0.2344,
+      "step": 2682
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6880861830302394e-05,
+      "loss": 0.7655,
+      "step": 2683
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6812916632878066e-05,
+      "loss": 0.7733,
+      "step": 2684
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.674501995868429e-05,
+      "loss": 0.7956,
+      "step": 2685
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.667717185986085e-05,
+      "loss": 0.8116,
+      "step": 2686
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.660937238851034e-05,
+      "loss": 0.836,
+      "step": 2687
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.654162159669802e-05,
+      "loss": 0.8366,
+      "step": 2688
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.647391953645164e-05,
+      "loss": 0.8201,
+      "step": 2689
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.640626625976168e-05,
+      "loss": 0.8146,
+      "step": 2690
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.63386618185811e-05,
+      "loss": 0.7649,
+      "step": 2691
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6271106264825294e-05,
+      "loss": 0.7375,
+      "step": 2692
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.620359965037221e-05,
+      "loss": 0.7921,
+      "step": 2693
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.613614202706219e-05,
+      "loss": 0.8362,
+      "step": 2694
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.606873344669789e-05,
+      "loss": 0.7706,
+      "step": 2695
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6001373961044385e-05,
+      "loss": 0.7554,
+      "step": 2696
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.593406362182903e-05,
+      "loss": 0.775,
+      "step": 2697
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5866802480741415e-05,
+      "loss": 0.829,
+      "step": 2698
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.579959058943331e-05,
+      "loss": 0.8391,
+      "step": 2699
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.573242799951876e-05,
+      "loss": 0.7823,
+      "step": 2700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5665314762573934e-05,
+      "loss": 0.8159,
+      "step": 2701
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5598250930137e-05,
+      "loss": 0.8046,
+      "step": 2702
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5531236553708305e-05,
+      "loss": 0.7675,
+      "step": 2703
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.546427168475018e-05,
+      "loss": 0.8161,
+      "step": 2704
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5397356374686895e-05,
+      "loss": 0.781,
+      "step": 2705
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5330490674904735e-05,
+      "loss": 0.81,
+      "step": 2706
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5263674636751795e-05,
+      "loss": 0.8754,
+      "step": 2707
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5196908311538125e-05,
+      "loss": 0.8116,
+      "step": 2708
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.513019175053559e-05,
+      "loss": 0.8008,
+      "step": 2709
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5063525004977724e-05,
+      "loss": 0.8008,
+      "step": 2710
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.499690812605997e-05,
+      "loss": 0.793,
+      "step": 2711
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.49303411649394e-05,
+      "loss": 0.8132,
+      "step": 2712
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4863824172734686e-05,
+      "loss": 0.8075,
+      "step": 2713
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.479735720052622e-05,
+      "loss": 0.7886,
+      "step": 2714
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.473094029935603e-05,
+      "loss": 0.7945,
+      "step": 2715
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4664573520227564e-05,
+      "loss": 0.2394,
+      "step": 2716
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4598256914105756e-05,
+      "loss": 0.7998,
+      "step": 2717
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.453199053191722e-05,
+      "loss": 0.7882,
+      "step": 2718
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4465774424549834e-05,
+      "loss": 0.8415,
+      "step": 2719
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.439960864285281e-05,
+      "loss": 0.7367,
+      "step": 2720
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.433349323763696e-05,
+      "loss": 0.801,
+      "step": 2721
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4267428259674186e-05,
+      "loss": 0.2539,
+      "step": 2722
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.42014137596977e-05,
+      "loss": 0.8206,
+      "step": 2723
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4135449788402044e-05,
+      "loss": 0.8252,
+      "step": 2724
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.40695363964429e-05,
+      "loss": 0.7987,
+      "step": 2725
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4003673634437083e-05,
+      "loss": 0.7984,
+      "step": 2726
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3937861552962555e-05,
+      "loss": 0.8478,
+      "step": 2727
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.387210020255843e-05,
+      "loss": 0.8141,
+      "step": 2728
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.38063896337247e-05,
+      "loss": 0.849,
+      "step": 2729
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3740729896922496e-05,
+      "loss": 0.7837,
+      "step": 2730
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.367512104257391e-05,
+      "loss": 0.8124,
+      "step": 2731
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.360956312106185e-05,
+      "loss": 0.7975,
+      "step": 2732
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3544056182730244e-05,
+      "loss": 0.823,
+      "step": 2733
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3478600277883756e-05,
+      "loss": 0.8002,
+      "step": 2734
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3413195456787935e-05,
+      "loss": 0.7985,
+      "step": 2735
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.334784176966912e-05,
+      "loss": 0.7525,
+      "step": 2736
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.328253926671426e-05,
+      "loss": 0.7639,
+      "step": 2737
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.321728799807112e-05,
+      "loss": 0.8122,
+      "step": 2738
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.315208801384811e-05,
+      "loss": 0.8123,
+      "step": 2739
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.308693936411421e-05,
+      "loss": 0.793,
+      "step": 2740
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3021842098898934e-05,
+      "loss": 0.7962,
+      "step": 2741
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2956796268192456e-05,
+      "loss": 0.8301,
+      "step": 2742
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.289180192194541e-05,
+      "loss": 0.8008,
+      "step": 2743
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.282685911006882e-05,
+      "loss": 0.2255,
+      "step": 2744
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.276196788243422e-05,
+      "loss": 0.7848,
+      "step": 2745
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.269712828887354e-05,
+      "loss": 0.8336,
+      "step": 2746
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.263234037917894e-05,
+      "loss": 0.8204,
+      "step": 2747
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.256760420310303e-05,
+      "loss": 0.8332,
+      "step": 2748
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2502919810358666e-05,
+      "loss": 0.8095,
+      "step": 2749
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.243828725061882e-05,
+      "loss": 0.8552,
+      "step": 2750
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2373706573516794e-05,
+      "loss": 0.8353,
+      "step": 2751
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.230917782864603e-05,
+      "loss": 0.7813,
+      "step": 2752
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.224470106556e-05,
+      "loss": 0.8281,
+      "step": 2753
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.218027633377235e-05,
+      "loss": 0.8269,
+      "step": 2754
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.211590368275675e-05,
+      "loss": 0.7904,
+      "step": 2755
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.205158316194686e-05,
+      "loss": 0.8118,
+      "step": 2756
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1987314820736257e-05,
+      "loss": 0.8072,
+      "step": 2757
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.192309870847855e-05,
+      "loss": 0.8057,
+      "step": 2758
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.18589348744872e-05,
+      "loss": 0.8018,
+      "step": 2759
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1794823368035476e-05,
+      "loss": 0.7814,
+      "step": 2760
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1730764238356514e-05,
+      "loss": 0.8037,
+      "step": 2761
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1666757534643255e-05,
+      "loss": 0.8341,
+      "step": 2762
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.160280330604827e-05,
+      "loss": 0.7678,
+      "step": 2763
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1538901601683945e-05,
+      "loss": 0.7799,
+      "step": 2764
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1475052470622315e-05,
+      "loss": 0.7821,
+      "step": 2765
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.141125596189494e-05,
+      "loss": 0.8404,
+      "step": 2766
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1347512124493125e-05,
+      "loss": 0.2363,
+      "step": 2767
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.12838210073676e-05,
+      "loss": 0.8003,
+      "step": 2768
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1220182659428655e-05,
+      "loss": 0.8078,
+      "step": 2769
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.115659712954612e-05,
+      "loss": 0.8395,
+      "step": 2770
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.109306446654912e-05,
+      "loss": 0.8246,
+      "step": 2771
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.10295847192263e-05,
+      "loss": 0.7982,
+      "step": 2772
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.096615793632569e-05,
+      "loss": 0.2392,
+      "step": 2773
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.090278416655454e-05,
+      "loss": 0.7924,
+      "step": 2774
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.083946345857939e-05,
+      "loss": 0.8198,
+      "step": 2775
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.07761958610262e-05,
+      "loss": 0.7833,
+      "step": 2776
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.071298142247998e-05,
+      "loss": 0.8039,
+      "step": 2777
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0649820191484904e-05,
+      "loss": 0.7761,
+      "step": 2778
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.058671221654446e-05,
+      "loss": 0.8279,
+      "step": 2779
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.052365754612107e-05,
+      "loss": 0.7988,
+      "step": 2780
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0460656228636254e-05,
+      "loss": 0.8308,
+      "step": 2781
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.039770831247063e-05,
+      "loss": 0.8128,
+      "step": 2782
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0334813845963784e-05,
+      "loss": 0.8116,
+      "step": 2783
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0271972877414167e-05,
+      "loss": 0.7712,
+      "step": 2784
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.020918545507926e-05,
+      "loss": 0.8178,
+      "step": 2785
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.014645162717542e-05,
+      "loss": 0.8209,
+      "step": 2786
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0083771441877728e-05,
+      "loss": 0.8148,
+      "step": 2787
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.00211449473202e-05,
+      "loss": 0.7888,
+      "step": 2788
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.99585721915956e-05,
+      "loss": 0.7944,
+      "step": 2789
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9896053222755338e-05,
+      "loss": 0.7475,
+      "step": 2790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9833588088809628e-05,
+      "loss": 0.8137,
+      "step": 2791
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9771176837727255e-05,
+      "loss": 0.7836,
+      "step": 2792
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.970881951743567e-05,
+      "loss": 0.8476,
+      "step": 2793
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.964651617582096e-05,
+      "loss": 0.8213,
+      "step": 2794
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9584266860727628e-05,
+      "loss": 0.795,
+      "step": 2795
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.952207161995879e-05,
+      "loss": 0.7931,
+      "step": 2796
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9459930501276058e-05,
+      "loss": 0.808,
+      "step": 2797
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9397843552399397e-05,
+      "loss": 0.8159,
+      "step": 2798
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.933581082100717e-05,
+      "loss": 0.8061,
+      "step": 2799
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9273832354736197e-05,
+      "loss": 0.7858,
+      "step": 2800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9211908201181593e-05,
+      "loss": 0.7816,
+      "step": 2801
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9150038407896696e-05,
+      "loss": 0.8043,
+      "step": 2802
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.908822302239317e-05,
+      "loss": 0.795,
+      "step": 2803
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9026462092140915e-05,
+      "loss": 0.803,
+      "step": 2804
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8964755664567923e-05,
+      "loss": 0.7962,
+      "step": 2805
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.890310378706039e-05,
+      "loss": 0.8113,
+      "step": 2806
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8841506506962678e-05,
+      "loss": 0.833,
+      "step": 2807
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8779963871577076e-05,
+      "loss": 0.8108,
+      "step": 2808
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8718475928164034e-05,
+      "loss": 0.7956,
+      "step": 2809
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8657042723941996e-05,
+      "loss": 0.8302,
+      "step": 2810
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8595664306087312e-05,
+      "loss": 0.8405,
+      "step": 2811
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.853434072173421e-05,
+      "loss": 0.7902,
+      "step": 2812
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8473072017975012e-05,
+      "loss": 0.2423,
+      "step": 2813
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.84118582418597e-05,
+      "loss": 0.8122,
+      "step": 2814
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8350699440396123e-05,
+      "loss": 0.2393,
+      "step": 2815
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.828959566054996e-05,
+      "loss": 0.8249,
+      "step": 2816
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.822854694924465e-05,
+      "loss": 0.8164,
+      "step": 2817
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.816755335336123e-05,
+      "loss": 0.8068,
+      "step": 2818
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8106614919738526e-05,
+      "loss": 0.7366,
+      "step": 2819
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8045731695172994e-05,
+      "loss": 0.8067,
+      "step": 2820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7984903726418598e-05,
+      "loss": 0.7723,
+      "step": 2821
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7924131060186963e-05,
+      "loss": 0.7914,
+      "step": 2822
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.786341374314725e-05,
+      "loss": 0.8034,
+      "step": 2823
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.780275182192603e-05,
+      "loss": 0.7724,
+      "step": 2824
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.774214534310744e-05,
+      "loss": 0.7832,
+      "step": 2825
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7681594353232932e-05,
+      "loss": 0.8235,
+      "step": 2826
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7621098898801422e-05,
+      "loss": 0.7967,
+      "step": 2827
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.75606590262692e-05,
+      "loss": 0.7515,
+      "step": 2828
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7500274782049762e-05,
+      "loss": 0.8317,
+      "step": 2829
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7439946212514e-05,
+      "loss": 0.8127,
+      "step": 2830
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.737967336399002e-05,
+      "loss": 0.2398,
+      "step": 2831
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7319456282763113e-05,
+      "loss": 0.8312,
+      "step": 2832
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.725929501507568e-05,
+      "loss": 0.8119,
+      "step": 2833
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7199189607127472e-05,
+      "loss": 0.7877,
+      "step": 2834
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7139140105075134e-05,
+      "loss": 0.8484,
+      "step": 2835
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.707914655503242e-05,
+      "loss": 0.8053,
+      "step": 2836
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7019209003070245e-05,
+      "loss": 0.7563,
+      "step": 2837
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.695932749521638e-05,
+      "loss": 0.8412,
+      "step": 2838
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6899502077455584e-05,
+      "loss": 0.7883,
+      "step": 2839
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.683973279572959e-05,
+      "loss": 0.8061,
+      "step": 2840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6780019695937008e-05,
+      "loss": 0.8124,
+      "step": 2841
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6720362823933266e-05,
+      "loss": 0.8091,
+      "step": 2842
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6660762225530657e-05,
+      "loss": 0.2347,
+      "step": 2843
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.660121794649826e-05,
+      "loss": 0.8022,
+      "step": 2844
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.654173003256184e-05,
+      "loss": 0.787,
+      "step": 2845
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.648229852940397e-05,
+      "loss": 0.7984,
+      "step": 2846
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6422923482663863e-05,
+      "loss": 0.7645,
+      "step": 2847
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6363604937937337e-05,
+      "loss": 0.7711,
+      "step": 2848
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6304342940776895e-05,
+      "loss": 0.8088,
+      "step": 2849
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6245137536691532e-05,
+      "loss": 0.7809,
+      "step": 2850
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6185988771146863e-05,
+      "loss": 0.806,
+      "step": 2851
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6126896689564994e-05,
+      "loss": 0.7678,
+      "step": 2852
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.606786133732442e-05,
+      "loss": 0.2352,
+      "step": 2853
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.600888275976021e-05,
+      "loss": 0.8057,
+      "step": 2854
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5949961002163692e-05,
+      "loss": 0.2538,
+      "step": 2855
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5891096109782642e-05,
+      "loss": 0.7808,
+      "step": 2856
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5832288127821202e-05,
+      "loss": 0.819,
+      "step": 2857
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5773537101439694e-05,
+      "loss": 0.8018,
+      "step": 2858
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5714843075754823e-05,
+      "loss": 0.8149,
+      "step": 2859
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.565620609583942e-05,
+      "loss": 0.817,
+      "step": 2860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.559762620672258e-05,
+      "loss": 0.7915,
+      "step": 2861
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5539103453389556e-05,
+      "loss": 0.8302,
+      "step": 2862
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5480637880781656e-05,
+      "loss": 0.8282,
+      "step": 2863
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.542222953379634e-05,
+      "loss": 0.8127,
+      "step": 2864
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5363878457287138e-05,
+      "loss": 0.7836,
+      "step": 2865
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.530558469606351e-05,
+      "loss": 0.7803,
+      "step": 2866
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5247348294890983e-05,
+      "loss": 0.8044,
+      "step": 2867
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5189169298491054e-05,
+      "loss": 0.8218,
+      "step": 2868
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.513104775154107e-05,
+      "loss": 0.8232,
+      "step": 2869
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.50729836986742e-05,
+      "loss": 0.7722,
+      "step": 2870
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5014977184479694e-05,
+      "loss": 0.7719,
+      "step": 2871
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.49570282535024e-05,
+      "loss": 0.8545,
+      "step": 2872
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4899136950242986e-05,
+      "loss": 0.7672,
+      "step": 2873
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4841303319157925e-05,
+      "loss": 0.7844,
+      "step": 2874
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.47835274046594e-05,
+      "loss": 0.8108,
+      "step": 2875
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4725809251115207e-05,
+      "loss": 0.7885,
+      "step": 2876
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.466814890284882e-05,
+      "loss": 0.7751,
+      "step": 2877
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.461054640413938e-05,
+      "loss": 0.7949,
+      "step": 2878
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4553001799221474e-05,
+      "loss": 0.8153,
+      "step": 2879
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.449551513228535e-05,
+      "loss": 0.7729,
+      "step": 2880
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.443808644747675e-05,
+      "loss": 0.7907,
+      "step": 2881
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.438071578889679e-05,
+      "loss": 0.8214,
+      "step": 2882
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.432340320060216e-05,
+      "loss": 0.8154,
+      "step": 2883
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.426614872660483e-05,
+      "loss": 0.2097,
+      "step": 2884
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4208952410872244e-05,
+      "loss": 0.8218,
+      "step": 2885
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4151814297327158e-05,
+      "loss": 0.8081,
+      "step": 2886
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4094734429847575e-05,
+      "loss": 0.7955,
+      "step": 2887
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4037712852266836e-05,
+      "loss": 0.7975,
+      "step": 2888
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3980749608373533e-05,
+      "loss": 0.7827,
+      "step": 2889
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3923844741911385e-05,
+      "loss": 0.8308,
+      "step": 2890
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.386699829657928e-05,
+      "loss": 0.2416,
+      "step": 2891
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.381021031603139e-05,
+      "loss": 0.7718,
+      "step": 2892
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.375348084387682e-05,
+      "loss": 0.779,
+      "step": 2893
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3696809923679774e-05,
+      "loss": 0.2161,
+      "step": 2894
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3640197598959622e-05,
+      "loss": 0.8156,
+      "step": 2895
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.35836439131906e-05,
+      "loss": 0.2201,
+      "step": 2896
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3527148909801922e-05,
+      "loss": 0.8149,
+      "step": 2897
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3470712632177804e-05,
+      "loss": 0.8259,
+      "step": 2898
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.341433512365735e-05,
+      "loss": 0.7742,
+      "step": 2899
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.335801642753448e-05,
+      "loss": 0.7841,
+      "step": 2900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3301756587057987e-05,
+      "loss": 0.8019,
+      "step": 2901
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3245555645431527e-05,
+      "loss": 0.8281,
+      "step": 2902
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3189413645813383e-05,
+      "loss": 0.8483,
+      "step": 2903
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3133330631316708e-05,
+      "loss": 0.7903,
+      "step": 2904
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.307730664500931e-05,
+      "loss": 0.7865,
+      "step": 2905
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3021341729913625e-05,
+      "loss": 0.774,
+      "step": 2906
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.296543592900683e-05,
+      "loss": 0.7487,
+      "step": 2907
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.290958928522058e-05,
+      "loss": 0.228,
+      "step": 2908
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.285380184144118e-05,
+      "loss": 0.7851,
+      "step": 2909
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.27980736405095e-05,
+      "loss": 0.7782,
+      "step": 2910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2742404725220823e-05,
+      "loss": 0.7868,
+      "step": 2911
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2686795138325e-05,
+      "loss": 0.2437,
+      "step": 2912
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.263124492252623e-05,
+      "loss": 0.7981,
+      "step": 2913
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2575754120483182e-05,
+      "loss": 0.738,
+      "step": 2914
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2520322774808933e-05,
+      "loss": 0.7577,
+      "step": 2915
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.246495092807077e-05,
+      "loss": 0.7737,
+      "step": 2916
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.240963862279046e-05,
+      "loss": 0.7641,
+      "step": 2917
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2354385901443887e-05,
+      "loss": 0.8422,
+      "step": 2918
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2299192806461268e-05,
+      "loss": 0.7878,
+      "step": 2919
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2244059380227056e-05,
+      "loss": 0.7833,
+      "step": 2920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2188985665079777e-05,
+      "loss": 0.7979,
+      "step": 2921
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2133971703312194e-05,
+      "loss": 0.7646,
+      "step": 2922
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2079017537171188e-05,
+      "loss": 0.8036,
+      "step": 2923
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2024123208857628e-05,
+      "loss": 0.7914,
+      "step": 2924
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.196928876052652e-05,
+      "loss": 0.784,
+      "step": 2925
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1914514234286897e-05,
+      "loss": 0.8059,
+      "step": 2926
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.18597996722017e-05,
+      "loss": 0.7955,
+      "step": 2927
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1805145116287807e-05,
+      "loss": 0.799,
+      "step": 2928
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1750550608516185e-05,
+      "loss": 0.8012,
+      "step": 2929
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1696016190811496e-05,
+      "loss": 0.8051,
+      "step": 2930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.164154190505231e-05,
+      "loss": 0.7888,
+      "step": 2931
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1587127793071127e-05,
+      "loss": 0.7877,
+      "step": 2932
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.15327738966541e-05,
+      "loss": 0.7769,
+      "step": 2933
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1478480257541177e-05,
+      "loss": 0.7781,
+      "step": 2934
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.142424691742607e-05,
+      "loss": 0.8036,
+      "step": 2935
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1370073917956202e-05,
+      "loss": 0.8052,
+      "step": 2936
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1315961300732557e-05,
+      "loss": 0.8228,
+      "step": 2937
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.126190910730984e-05,
+      "loss": 0.2529,
+      "step": 2938
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1207917379196373e-05,
+      "loss": 0.8108,
+      "step": 2939
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1153986157853923e-05,
+      "loss": 0.8118,
+      "step": 2940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1100115484697947e-05,
+      "loss": 0.7957,
+      "step": 2941
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1046305401097255e-05,
+      "loss": 0.8214,
+      "step": 2942
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.099255594837424e-05,
+      "loss": 0.7659,
+      "step": 2943
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.093886716780472e-05,
+      "loss": 0.8085,
+      "step": 2944
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.088523910061785e-05,
+      "loss": 0.827,
+      "step": 2945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.083167178799623e-05,
+      "loss": 0.8061,
+      "step": 2946
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0778165271075823e-05,
+      "loss": 0.7972,
+      "step": 2947
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0724719590945784e-05,
+      "loss": 0.251,
+      "step": 2948
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.067133478864869e-05,
+      "loss": 0.7886,
+      "step": 2949
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0618010905180318e-05,
+      "loss": 0.8059,
+      "step": 2950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.056474798148964e-05,
+      "loss": 0.8044,
+      "step": 2951
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0511546058478793e-05,
+      "loss": 0.7872,
+      "step": 2952
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0458405177003138e-05,
+      "loss": 0.8042,
+      "step": 2953
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0405325377871163e-05,
+      "loss": 0.8492,
+      "step": 2954
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.035230670184436e-05,
+      "loss": 0.7879,
+      "step": 2955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0299349189637353e-05,
+      "loss": 0.8069,
+      "step": 2956
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0246452881917822e-05,
+      "loss": 0.7849,
+      "step": 2957
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0193617819306353e-05,
+      "loss": 0.7777,
+      "step": 2958
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0140844042376583e-05,
+      "loss": 0.2205,
+      "step": 2959
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0088131591655068e-05,
+      "loss": 0.8176,
+      "step": 2960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0035480507621218e-05,
+      "loss": 0.8206,
+      "step": 2961
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.998289083070738e-05,
+      "loss": 0.8229,
+      "step": 2962
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9930362601298746e-05,
+      "loss": 0.7438,
+      "step": 2963
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9877895859733252e-05,
+      "loss": 0.2442,
+      "step": 2964
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9825490646301715e-05,
+      "loss": 0.7715,
+      "step": 2965
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9773147001247582e-05,
+      "loss": 0.8377,
+      "step": 2966
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.972086496476716e-05,
+      "loss": 0.7765,
+      "step": 2967
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.96686445770093e-05,
+      "loss": 0.7382,
+      "step": 2968
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9616485878075607e-05,
+      "loss": 0.8128,
+      "step": 2969
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.956438890802034e-05,
+      "loss": 0.746,
+      "step": 2970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.951235370685023e-05,
+      "loss": 0.827,
+      "step": 2971
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9460380314524675e-05,
+      "loss": 0.8091,
+      "step": 2972
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.94084687709556e-05,
+      "loss": 0.8107,
+      "step": 2973
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9356619116007392e-05,
+      "loss": 0.7925,
+      "step": 2974
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9304831389496957e-05,
+      "loss": 0.7904,
+      "step": 2975
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.925310563119358e-05,
+      "loss": 0.7882,
+      "step": 2976
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9201441880819026e-05,
+      "loss": 0.8067,
+      "step": 2977
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.914984017804745e-05,
+      "loss": 0.7659,
+      "step": 2978
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9098300562505266e-05,
+      "loss": 0.8075,
+      "step": 2979
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9046823073771293e-05,
+      "loss": 0.8307,
+      "step": 2980
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.899540775137666e-05,
+      "loss": 0.7887,
+      "step": 2981
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8944054634804666e-05,
+      "loss": 0.7951,
+      "step": 2982
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8892763763490852e-05,
+      "loss": 0.8136,
+      "step": 2983
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.88415351768231e-05,
+      "loss": 0.7932,
+      "step": 2984
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.879036891414129e-05,
+      "loss": 0.8313,
+      "step": 2985
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.873926501473747e-05,
+      "loss": 0.7878,
+      "step": 2986
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8688223517855942e-05,
+      "loss": 0.8094,
+      "step": 2987
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8637244462692915e-05,
+      "loss": 0.7959,
+      "step": 2988
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8586327888396672e-05,
+      "loss": 0.2417,
+      "step": 2989
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8535473834067663e-05,
+      "loss": 0.2126,
+      "step": 2990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8484682338758152e-05,
+      "loss": 0.7782,
+      "step": 2991
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.843395344147243e-05,
+      "loss": 0.8036,
+      "step": 2992
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8383287181166732e-05,
+      "loss": 0.7797,
+      "step": 2993
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.833268359674919e-05,
+      "loss": 0.7648,
+      "step": 2994
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8282142727079764e-05,
+      "loss": 0.7856,
+      "step": 2995
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.82316646109703e-05,
+      "loss": 0.8425,
+      "step": 2996
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8181249287184466e-05,
+      "loss": 0.7712,
+      "step": 2997
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8130896794437614e-05,
+      "loss": 0.8209,
+      "step": 2998
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8080607171396978e-05,
+      "loss": 0.8174,
+      "step": 2999
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8030380456681396e-05,
+      "loss": 0.775,
+      "step": 3000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7980216688861472e-05,
+      "loss": 0.8541,
+      "step": 3001
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7930115906459465e-05,
+      "loss": 0.8326,
+      "step": 3002
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7880078147949207e-05,
+      "loss": 0.7893,
+      "step": 3003
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.78301034517562e-05,
+      "loss": 0.7663,
+      "step": 3004
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.77801918562575e-05,
+      "loss": 0.8218,
+      "step": 3005
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7730343399781668e-05,
+      "loss": 0.8126,
+      "step": 3006
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7680558120608825e-05,
+      "loss": 0.7816,
+      "step": 3007
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7630836056970578e-05,
+      "loss": 0.8012,
+      "step": 3008
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.758117724704995e-05,
+      "loss": 0.8074,
+      "step": 3009
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7531581728981384e-05,
+      "loss": 0.8047,
+      "step": 3010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7482049540850788e-05,
+      "loss": 0.8286,
+      "step": 3011
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.743258072069539e-05,
+      "loss": 0.7715,
+      "step": 3012
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.738317530650373e-05,
+      "loss": 0.8182,
+      "step": 3013
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.73338333362157e-05,
+      "loss": 0.8036,
+      "step": 3014
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.72845548477225e-05,
+      "loss": 0.7943,
+      "step": 3015
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7235339878866476e-05,
+      "loss": 0.7638,
+      "step": 3016
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.718618846744129e-05,
+      "loss": 0.802,
+      "step": 3017
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7137100651191773e-05,
+      "loss": 0.8076,
+      "step": 3018
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7088076467813885e-05,
+      "loss": 0.7787,
+      "step": 3019
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7039115954954765e-05,
+      "loss": 0.7883,
+      "step": 3020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.699021915021266e-05,
+      "loss": 0.794,
+      "step": 3021
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6941386091136824e-05,
+      "loss": 0.8271,
+      "step": 3022
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6892616815227624e-05,
+      "loss": 0.7946,
+      "step": 3023
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.684391135993647e-05,
+      "loss": 0.8345,
+      "step": 3024
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6795269762665688e-05,
+      "loss": 0.2432,
+      "step": 3025
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6746692060768578e-05,
+      "loss": 0.8108,
+      "step": 3026
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6698178291549417e-05,
+      "loss": 0.8006,
+      "step": 3027
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6649728492263383e-05,
+      "loss": 0.8471,
+      "step": 3028
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6601342700116473e-05,
+      "loss": 0.8042,
+      "step": 3029
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6553020952265587e-05,
+      "loss": 0.7752,
+      "step": 3030
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6504763285818438e-05,
+      "loss": 0.2305,
+      "step": 3031
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6456569737833483e-05,
+      "loss": 0.8197,
+      "step": 3032
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.640844034532002e-05,
+      "loss": 0.7976,
+      "step": 3033
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6360375145237983e-05,
+      "loss": 0.8066,
+      "step": 3034
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.631237417449808e-05,
+      "loss": 0.8061,
+      "step": 3035
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6264437469961703e-05,
+      "loss": 0.7718,
+      "step": 3036
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6216565068440815e-05,
+      "loss": 0.7765,
+      "step": 3037
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6168757006698075e-05,
+      "loss": 0.8089,
+      "step": 3038
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6121013321446733e-05,
+      "loss": 0.7754,
+      "step": 3039
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.607333404935053e-05,
+      "loss": 0.8268,
+      "step": 3040
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6025719227023817e-05,
+      "loss": 0.7768,
+      "step": 3041
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.597816889103144e-05,
+      "loss": 0.8226,
+      "step": 3042
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5930683077888688e-05,
+      "loss": 0.8461,
+      "step": 3043
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.588326182406128e-05,
+      "loss": 0.8099,
+      "step": 3044
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.583590516596548e-05,
+      "loss": 0.8011,
+      "step": 3045
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5788613139967846e-05,
+      "loss": 0.7592,
+      "step": 3046
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.574138578238524e-05,
+      "loss": 0.7897,
+      "step": 3047
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5694223129485075e-05,
+      "loss": 0.8268,
+      "step": 3048
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.564712521748488e-05,
+      "loss": 0.776,
+      "step": 3049
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.560009208255251e-05,
+      "loss": 0.8338,
+      "step": 3050
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5553123760806143e-05,
+      "loss": 0.8157,
+      "step": 3051
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5506220288314144e-05,
+      "loss": 0.8107,
+      "step": 3052
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5459381701095053e-05,
+      "loss": 0.8042,
+      "step": 3053
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5412608035117604e-05,
+      "loss": 0.8137,
+      "step": 3054
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5365899326300726e-05,
+      "loss": 0.796,
+      "step": 3055
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.531925561051336e-05,
+      "loss": 0.7923,
+      "step": 3056
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5272676923574637e-05,
+      "loss": 0.7954,
+      "step": 3057
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.522616330125367e-05,
+      "loss": 0.7694,
+      "step": 3058
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.517971477926967e-05,
+      "loss": 0.8222,
+      "step": 3059
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.513333139329186e-05,
+      "loss": 0.8727,
+      "step": 3060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5087013178939346e-05,
+      "loss": 0.7995,
+      "step": 3061
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5040760171781298e-05,
+      "loss": 0.8612,
+      "step": 3062
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4994572407336783e-05,
+      "loss": 0.8173,
+      "step": 3063
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4948449921074702e-05,
+      "loss": 0.792,
+      "step": 3064
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4902392748413907e-05,
+      "loss": 0.783,
+      "step": 3065
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4856400924723079e-05,
+      "loss": 0.8051,
+      "step": 3066
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.481047448532069e-05,
+      "loss": 0.7735,
+      "step": 3067
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4764613465474986e-05,
+      "loss": 0.8242,
+      "step": 3068
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4718817900404014e-05,
+      "loss": 0.7717,
+      "step": 3069
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4673087825275567e-05,
+      "loss": 0.7676,
+      "step": 3070
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4627423275207076e-05,
+      "loss": 0.8101,
+      "step": 3071
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4581824285265733e-05,
+      "loss": 0.778,
+      "step": 3072
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4536290890468362e-05,
+      "loss": 0.8366,
+      "step": 3073
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4490823125781372e-05,
+      "loss": 0.7849,
+      "step": 3074
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.444542102612082e-05,
+      "loss": 0.8272,
+      "step": 3075
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4400084626352339e-05,
+      "loss": 0.8243,
+      "step": 3076
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4354813961291047e-05,
+      "loss": 0.805,
+      "step": 3077
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4309609065701656e-05,
+      "loss": 0.8132,
+      "step": 3078
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4264469974298367e-05,
+      "loss": 0.7878,
+      "step": 3079
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4219396721744793e-05,
+      "loss": 0.8083,
+      "step": 3080
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4174389342653971e-05,
+      "loss": 0.8127,
+      "step": 3081
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4129447871588508e-05,
+      "loss": 0.7946,
+      "step": 3082
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.408457234306022e-05,
+      "loss": 0.8268,
+      "step": 3083
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4039762791530343e-05,
+      "loss": 0.8215,
+      "step": 3084
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.399501925140947e-05,
+      "loss": 0.7814,
+      "step": 3085
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3950341757057528e-05,
+      "loss": 0.7915,
+      "step": 3086
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3905730342783629e-05,
+      "loss": 0.2419,
+      "step": 3087
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3861185042846226e-05,
+      "loss": 0.792,
+      "step": 3088
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3816705891452986e-05,
+      "loss": 0.8155,
+      "step": 3089
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3772292922760732e-05,
+      "loss": 0.7901,
+      "step": 3090
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3727946170875538e-05,
+      "loss": 0.8031,
+      "step": 3091
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.368366566985253e-05,
+      "loss": 0.7702,
+      "step": 3092
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.363945145369606e-05,
+      "loss": 0.8076,
+      "step": 3093
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3595303556359539e-05,
+      "loss": 0.7814,
+      "step": 3094
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3551222011745391e-05,
+      "loss": 0.7848,
+      "step": 3095
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3507206853705178e-05,
+      "loss": 0.7782,
+      "step": 3096
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3463258116039457e-05,
+      "loss": 0.7833,
+      "step": 3097
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3419375832497727e-05,
+      "loss": 0.8348,
+      "step": 3098
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3375560036778501e-05,
+      "loss": 0.7832,
+      "step": 3099
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.333181076252924e-05,
+      "loss": 0.8036,
+      "step": 3100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3288128043346315e-05,
+      "loss": 0.7633,
+      "step": 3101
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3244511912774905e-05,
+      "loss": 0.8437,
+      "step": 3102
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3200962404309225e-05,
+      "loss": 0.8018,
+      "step": 3103
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3157479551392182e-05,
+      "loss": 0.7762,
+      "step": 3104
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.311406338741551e-05,
+      "loss": 0.8018,
+      "step": 3105
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3070713945719859e-05,
+      "loss": 0.7872,
+      "step": 3106
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.302743125959448e-05,
+      "loss": 0.8182,
+      "step": 3107
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.298421536227743e-05,
+      "loss": 0.7444,
+      "step": 3108
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2941066286955483e-05,
+      "loss": 0.7958,
+      "step": 3109
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2897984066764113e-05,
+      "loss": 0.7852,
+      "step": 3110
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2854968734787398e-05,
+      "loss": 0.7655,
+      "step": 3111
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2812020324058094e-05,
+      "loss": 0.839,
+      "step": 3112
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2769138867557584e-05,
+      "loss": 0.8096,
+      "step": 3113
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2726324398215762e-05,
+      "loss": 0.8378,
+      "step": 3114
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2683576948911147e-05,
+      "loss": 0.7879,
+      "step": 3115
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2640896552470794e-05,
+      "loss": 0.7967,
+      "step": 3116
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2598283241670195e-05,
+      "loss": 0.7452,
+      "step": 3117
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2555737049233418e-05,
+      "loss": 0.7678,
+      "step": 3118
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2513258007832896e-05,
+      "loss": 0.8358,
+      "step": 3119
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2470846150089566e-05,
+      "loss": 0.7806,
+      "step": 3120
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2428501508572765e-05,
+      "loss": 0.7864,
+      "step": 3121
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.238622411580017e-05,
+      "loss": 0.8181,
+      "step": 3122
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2344014004237859e-05,
+      "loss": 0.8185,
+      "step": 3123
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2301871206300198e-05,
+      "loss": 0.8051,
+      "step": 3124
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2259795754349933e-05,
+      "loss": 0.7928,
+      "step": 3125
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.221778768069799e-05,
+      "loss": 0.7867,
+      "step": 3126
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2175847017603648e-05,
+      "loss": 0.7922,
+      "step": 3127
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.21339737972744e-05,
+      "loss": 0.7596,
+      "step": 3128
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2092168051865892e-05,
+      "loss": 0.8146,
+      "step": 3129
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2050429813482001e-05,
+      "loss": 0.8327,
+      "step": 3130
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2008759114174795e-05,
+      "loss": 0.8384,
+      "step": 3131
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1967155985944367e-05,
+      "loss": 0.8126,
+      "step": 3132
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.192562046073904e-05,
+      "loss": 0.8414,
+      "step": 3133
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1884152570455176e-05,
+      "loss": 0.716,
+      "step": 3134
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.184275234693717e-05,
+      "loss": 0.8014,
+      "step": 3135
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1801419821977478e-05,
+      "loss": 0.7973,
+      "step": 3136
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1760155027316589e-05,
+      "loss": 0.8153,
+      "step": 3137
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1718957994642965e-05,
+      "loss": 0.233,
+      "step": 3138
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1677828755592957e-05,
+      "loss": 0.7746,
+      "step": 3139
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1636767341751021e-05,
+      "loss": 0.7857,
+      "step": 3140
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1595773784649389e-05,
+      "loss": 0.7986,
+      "step": 3141
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1554848115768202e-05,
+      "loss": 0.7844,
+      "step": 3142
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.151399036653551e-05,
+      "loss": 0.8205,
+      "step": 3143
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1473200568327224e-05,
+      "loss": 0.7951,
+      "step": 3144
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1432478752466968e-05,
+      "loss": 0.8237,
+      "step": 3145
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.139182495022627e-05,
+      "loss": 0.8181,
+      "step": 3146
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1351239192824415e-05,
+      "loss": 0.798,
+      "step": 3147
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1310721511428335e-05,
+      "loss": 0.7414,
+      "step": 3148
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1270271937152831e-05,
+      "loss": 0.8058,
+      "step": 3149
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1229890501060258e-05,
+      "loss": 0.7804,
+      "step": 3150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1189577234160765e-05,
+      "loss": 0.832,
+      "step": 3151
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1149332167412097e-05,
+      "loss": 0.8158,
+      "step": 3152
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.11091553317196e-05,
+      "loss": 0.8014,
+      "step": 3153
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1069046757936275e-05,
+      "loss": 0.8004,
+      "step": 3154
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1029006476862702e-05,
+      "loss": 0.7819,
+      "step": 3155
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0989034519246954e-05,
+      "loss": 0.7716,
+      "step": 3156
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0949130915784711e-05,
+      "loss": 0.2311,
+      "step": 3157
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0909295697119148e-05,
+      "loss": 0.8172,
+      "step": 3158
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0869528893840875e-05,
+      "loss": 0.807,
+      "step": 3159
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0829830536487961e-05,
+      "loss": 0.7966,
+      "step": 3160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0790200655546034e-05,
+      "loss": 0.7849,
+      "step": 3161
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0750639281448017e-05,
+      "loss": 0.7785,
+      "step": 3162
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0711146444574205e-05,
+      "loss": 0.7854,
+      "step": 3163
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0671722175252408e-05,
+      "loss": 0.7723,
+      "step": 3164
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0632366503757652e-05,
+      "loss": 0.7784,
+      "step": 3165
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0593079460312282e-05,
+      "loss": 0.7806,
+      "step": 3166
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0553861075086014e-05,
+      "loss": 0.7837,
+      "step": 3167
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.051471137819583e-05,
+      "loss": 0.8316,
+      "step": 3168
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0475630399705894e-05,
+      "loss": 0.7332,
+      "step": 3169
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0436618169627655e-05,
+      "loss": 0.7964,
+      "step": 3170
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0397674717919802e-05,
+      "loss": 0.7892,
+      "step": 3171
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0358800074488106e-05,
+      "loss": 0.7908,
+      "step": 3172
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0319994269185585e-05,
+      "loss": 0.7362,
+      "step": 3173
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0281257331812377e-05,
+      "loss": 0.7533,
+      "step": 3174
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0242589292115689e-05,
+      "loss": 0.7804,
+      "step": 3175
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.020399017978989e-05,
+      "loss": 0.8021,
+      "step": 3176
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0165460024476337e-05,
+      "loss": 0.7275,
+      "step": 3177
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0126998855763503e-05,
+      "loss": 0.7863,
+      "step": 3178
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0088606703186876e-05,
+      "loss": 0.7698,
+      "step": 3179
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0050283596228871e-05,
+      "loss": 0.7541,
+      "step": 3180
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0012029564318992e-05,
+      "loss": 0.7748,
+      "step": 3181
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.973844636833606e-06,
+      "loss": 0.8028,
+      "step": 3182
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.93572884309607e-06,
+      "loss": 0.779,
+      "step": 3183
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.897682212376602e-06,
+      "loss": 0.7931,
+      "step": 3184
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.859704773892353e-06,
+      "loss": 0.7934,
+      "step": 3185
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.821796556807339e-06,
+      "loss": 0.814,
+      "step": 3186
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.783957590232362e-06,
+      "loss": 0.8052,
+      "step": 3187
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.7461879032251e-06,
+      "loss": 0.7361,
+      "step": 3188
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.708487524790045e-06,
+      "loss": 0.7348,
+      "step": 3189
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.670856483878388e-06,
+      "loss": 0.8074,
+      "step": 3190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.633294809388126e-06,
+      "loss": 0.8142,
+      "step": 3191
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.595802530164011e-06,
+      "loss": 0.8181,
+      "step": 3192
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.558379674997443e-06,
+      "loss": 0.7895,
+      "step": 3193
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.521026272626543e-06,
+      "loss": 0.745,
+      "step": 3194
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.483742351736113e-06,
+      "loss": 0.776,
+      "step": 3195
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.446527940957572e-06,
+      "loss": 0.7998,
+      "step": 3196
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.409383068868927e-06,
+      "loss": 0.7905,
+      "step": 3197
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.372307763994891e-06,
+      "loss": 0.788,
+      "step": 3198
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.33530205480665e-06,
+      "loss": 0.7858,
+      "step": 3199
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.298365969721968e-06,
+      "loss": 0.8233,
+      "step": 3200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.261499537105177e-06,
+      "loss": 0.793,
+      "step": 3201
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.224702785267114e-06,
+      "loss": 0.807,
+      "step": 3202
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.187975742465072e-06,
+      "loss": 0.8044,
+      "step": 3203
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.151318436902834e-06,
+      "loss": 0.7842,
+      "step": 3204
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.11473089673065e-06,
+      "loss": 0.8388,
+      "step": 3205
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.078213150045145e-06,
+      "loss": 0.78,
+      "step": 3206
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.041765224889399e-06,
+      "loss": 0.2434,
+      "step": 3207
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.005387149252853e-06,
+      "loss": 0.792,
+      "step": 3208
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.969078951071285e-06,
+      "loss": 0.7845,
+      "step": 3209
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.93284065822687e-06,
+      "loss": 0.8016,
+      "step": 3210
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.896672298548037e-06,
+      "loss": 0.8028,
+      "step": 3211
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.860573899809533e-06,
+      "loss": 0.8224,
+      "step": 3212
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.824545489732417e-06,
+      "loss": 0.7841,
+      "step": 3213
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.788587095983936e-06,
+      "loss": 0.8177,
+      "step": 3214
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.752698746177635e-06,
+      "loss": 0.7806,
+      "step": 3215
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.716880467873234e-06,
+      "loss": 0.7502,
+      "step": 3216
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.681132288576644e-06,
+      "loss": 0.7696,
+      "step": 3217
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.645454235739903e-06,
+      "loss": 0.7997,
+      "step": 3218
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.609846336761329e-06,
+      "loss": 0.7895,
+      "step": 3219
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.574308618985239e-06,
+      "loss": 0.7988,
+      "step": 3220
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.538841109702079e-06,
+      "loss": 0.7625,
+      "step": 3221
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.503443836148416e-06,
+      "loss": 0.805,
+      "step": 3222
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.468116825506877e-06,
+      "loss": 0.8161,
+      "step": 3223
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.432860104906093e-06,
+      "loss": 0.7909,
+      "step": 3224
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.39767370142076e-06,
+      "loss": 0.26,
+      "step": 3225
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.362557642071566e-06,
+      "loss": 0.795,
+      "step": 3226
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.32751195382514e-06,
+      "loss": 0.7622,
+      "step": 3227
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.29253666359413e-06,
+      "loss": 0.7931,
+      "step": 3228
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.257631798237097e-06,
+      "loss": 0.8012,
+      "step": 3229
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.2227973845585e-06,
+      "loss": 0.7719,
+      "step": 3230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.188033449308719e-06,
+      "loss": 0.8037,
+      "step": 3231
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.15334001918404e-06,
+      "loss": 0.8277,
+      "step": 3232
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.118717120826535e-06,
+      "loss": 0.7778,
+      "step": 3233
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.084164780824178e-06,
+      "loss": 0.7726,
+      "step": 3234
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.049683025710708e-06,
+      "loss": 0.7845,
+      "step": 3235
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.015271881965714e-06,
+      "loss": 0.7466,
+      "step": 3236
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.9809313760145e-06,
+      "loss": 0.8371,
+      "step": 3237
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.946661534228172e-06,
+      "loss": 0.7626,
+      "step": 3238
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.912462382923568e-06,
+      "loss": 0.7875,
+      "step": 3239
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.878333948363203e-06,
+      "loss": 0.2311,
+      "step": 3240
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.844276256755346e-06,
+      "loss": 0.8046,
+      "step": 3241
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.810289334253862e-06,
+      "loss": 0.803,
+      "step": 3242
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.776373206958332e-06,
+      "loss": 0.8007,
+      "step": 3243
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.742527900913976e-06,
+      "loss": 0.8245,
+      "step": 3244
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.708753442111572e-06,
+      "loss": 0.819,
+      "step": 3245
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.675049856487549e-06,
+      "loss": 0.8124,
+      "step": 3246
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.641417169923892e-06,
+      "loss": 0.7929,
+      "step": 3247
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.607855408248099e-06,
+      "loss": 0.7987,
+      "step": 3248
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.574364597233275e-06,
+      "loss": 0.8743,
+      "step": 3249
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.5409447625980015e-06,
+      "loss": 0.8669,
+      "step": 3250
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.50759593000635e-06,
+      "loss": 0.8252,
+      "step": 3251
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.4743181250678475e-06,
+      "loss": 0.7884,
+      "step": 3252
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.4411113733375726e-06,
+      "loss": 0.7976,
+      "step": 3253
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.407975700315928e-06,
+      "loss": 0.2369,
+      "step": 3254
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.374911131448747e-06,
+      "loss": 0.774,
+      "step": 3255
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.341917692127353e-06,
+      "loss": 0.7716,
+      "step": 3256
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.308995407688368e-06,
+      "loss": 0.8457,
+      "step": 3257
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.276144303413757e-06,
+      "loss": 0.7853,
+      "step": 3258
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.243364404530861e-06,
+      "loss": 0.8027,
+      "step": 3259
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.21065573621238e-06,
+      "loss": 0.8054,
+      "step": 3260
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.178018323576208e-06,
+      "loss": 0.7898,
+      "step": 3261
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.14545219168562e-06,
+      "loss": 0.8075,
+      "step": 3262
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.1129573655491e-06,
+      "loss": 0.8022,
+      "step": 3263
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.080533870120376e-06,
+      "loss": 0.7691,
+      "step": 3264
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.04818173029842e-06,
+      "loss": 0.8096,
+      "step": 3265
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.015900970927414e-06,
+      "loss": 0.7831,
+      "step": 3266
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.983691616796672e-06,
+      "loss": 0.7929,
+      "step": 3267
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.951553692640745e-06,
+      "loss": 0.7575,
+      "step": 3268
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.919487223139253e-06,
+      "loss": 0.7915,
+      "step": 3269
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.887492232917025e-06,
+      "loss": 0.8343,
+      "step": 3270
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.85556874654395e-06,
+      "loss": 0.7856,
+      "step": 3271
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.823716788535006e-06,
+      "loss": 0.7821,
+      "step": 3272
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.791936383350261e-06,
+      "loss": 0.7537,
+      "step": 3273
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.760227555394849e-06,
+      "loss": 0.8317,
+      "step": 3274
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.728590329018892e-06,
+      "loss": 0.8164,
+      "step": 3275
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.6970247285175315e-06,
+      "loss": 0.848,
+      "step": 3276
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.66553077813098e-06,
+      "loss": 0.7751,
+      "step": 3277
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.634108502044356e-06,
+      "loss": 0.8066,
+      "step": 3278
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.602757924387715e-06,
+      "loss": 0.761,
+      "step": 3279
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.571479069236131e-06,
+      "loss": 0.762,
+      "step": 3280
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.540271960609567e-06,
+      "loss": 0.2501,
+      "step": 3281
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.5091366224728554e-06,
+      "loss": 0.8144,
+      "step": 3282
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.478073078735758e-06,
+      "loss": 0.809,
+      "step": 3283
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.447081353252904e-06,
+      "loss": 0.8042,
+      "step": 3284
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.416161469823734e-06,
+      "loss": 0.7685,
+      "step": 3285
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.385313452192554e-06,
+      "loss": 0.8365,
+      "step": 3286
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.354537324048471e-06,
+      "loss": 0.7679,
+      "step": 3287
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.323833109025368e-06,
+      "loss": 0.826,
+      "step": 3288
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.2932008307019286e-06,
+      "loss": 0.8161,
+      "step": 3289
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.262640512601603e-06,
+      "loss": 0.8019,
+      "step": 3290
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.23215217819253e-06,
+      "loss": 0.8249,
+      "step": 3291
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.201735850887647e-06,
+      "loss": 0.8268,
+      "step": 3292
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.1713915540445035e-06,
+      "loss": 0.7697,
+      "step": 3293
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.1411193109654155e-06,
+      "loss": 0.8046,
+      "step": 3294
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.11091914489732e-06,
+      "loss": 0.8191,
+      "step": 3295
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.080791079031811e-06,
+      "loss": 0.7526,
+      "step": 3296
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.0507351365051475e-06,
+      "loss": 0.769,
+      "step": 3297
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.020751340398156e-06,
+      "loss": 0.8028,
+      "step": 3298
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.990839713736274e-06,
+      "loss": 0.8095,
+      "step": 3299
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.961000279489559e-06,
+      "loss": 0.7371,
+      "step": 3300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.9312330605725495e-06,
+      "loss": 0.2665,
+      "step": 3301
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.9015380798444265e-06,
+      "loss": 0.7739,
+      "step": 3302
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.871915360108781e-06,
+      "loss": 0.8034,
+      "step": 3303
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.842364924113819e-06,
+      "loss": 0.2467,
+      "step": 3304
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.812886794552197e-06,
+      "loss": 0.8351,
+      "step": 3305
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.7834809940610195e-06,
+      "loss": 0.2275,
+      "step": 3306
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.754147545221877e-06,
+      "loss": 0.7842,
+      "step": 3307
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.72488647056082e-06,
+      "loss": 0.745,
+      "step": 3308
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.69569779254826e-06,
+      "loss": 0.8307,
+      "step": 3309
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.666581533599025e-06,
+      "loss": 0.7792,
+      "step": 3310
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.637537716072416e-06,
+      "loss": 0.8437,
+      "step": 3311
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.608566362272017e-06,
+      "loss": 0.7928,
+      "step": 3312
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.579667494445739e-06,
+      "loss": 0.7912,
+      "step": 3313
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.550841134785945e-06,
+      "loss": 0.7793,
+      "step": 3314
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.522087305429214e-06,
+      "loss": 0.7864,
+      "step": 3315
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.49340602845646e-06,
+      "loss": 0.7276,
+      "step": 3316
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.46479732589289e-06,
+      "loss": 0.7801,
+      "step": 3317
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.436261219707994e-06,
+      "loss": 0.7802,
+      "step": 3318
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.407797731815456e-06,
+      "loss": 0.8028,
+      "step": 3319
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.379406884073246e-06,
+      "loss": 0.2336,
+      "step": 3320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.351088698283558e-06,
+      "loss": 0.8281,
+      "step": 3321
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.3228431961927285e-06,
+      "loss": 0.7986,
+      "step": 3322
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.294670399491319e-06,
+      "loss": 0.8144,
+      "step": 3323
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.266570329814091e-06,
+      "loss": 0.8001,
+      "step": 3324
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2385430087398715e-06,
+      "loss": 0.7583,
+      "step": 3325
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2105884577917095e-06,
+      "loss": 0.2476,
+      "step": 3326
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.18270669843669e-06,
+      "loss": 0.8128,
+      "step": 3327
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1548977520860745e-06,
+      "loss": 0.7803,
+      "step": 3328
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.127161640095168e-06,
+      "loss": 0.7522,
+      "step": 3329
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.099498383763346e-06,
+      "loss": 0.8113,
+      "step": 3330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.071908004334025e-06,
+      "loss": 0.7857,
+      "step": 3331
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.044390522994724e-06,
+      "loss": 0.8163,
+      "step": 3332
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.016945960876884e-06,
+      "loss": 0.8161,
+      "step": 3333
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.989574339055991e-06,
+      "loss": 0.7721,
+      "step": 3334
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.962275678551576e-06,
+      "loss": 0.7725,
+      "step": 3335
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9350500003270465e-06,
+      "loss": 0.8102,
+      "step": 3336
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.907897325289801e-06,
+      "loss": 0.8246,
+      "step": 3337
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.880817674291194e-06,
+      "loss": 0.7613,
+      "step": 3338
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.853811068126502e-06,
+      "loss": 0.7783,
+      "step": 3339
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.826877527534879e-06,
+      "loss": 0.7656,
+      "step": 3340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.800017073199404e-06,
+      "loss": 0.7877,
+      "step": 3341
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.773229725747008e-06,
+      "loss": 0.7947,
+      "step": 3342
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.746515505748472e-06,
+      "loss": 0.7761,
+      "step": 3343
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.719874433718452e-06,
+      "loss": 0.7564,
+      "step": 3344
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.693306530115415e-06,
+      "loss": 0.8042,
+      "step": 3345
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.666811815341643e-06,
+      "loss": 0.7769,
+      "step": 3346
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.640390309743193e-06,
+      "loss": 0.7788,
+      "step": 3347
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.614042033609945e-06,
+      "loss": 0.7767,
+      "step": 3348
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.587767007175503e-06,
+      "loss": 0.2384,
+      "step": 3349
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.56156525061725e-06,
+      "loss": 0.8099,
+      "step": 3350
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.535436784056269e-06,
+      "loss": 0.7719,
+      "step": 3351
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.509381627557419e-06,
+      "loss": 0.2329,
+      "step": 3352
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.483399801129173e-06,
+      "loss": 0.8115,
+      "step": 3353
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.457491324723784e-06,
+      "loss": 0.7641,
+      "step": 3354
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.431656218237135e-06,
+      "loss": 0.8067,
+      "step": 3355
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.405894501508745e-06,
+      "loss": 0.7889,
+      "step": 3356
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.380206194321812e-06,
+      "loss": 0.7504,
+      "step": 3357
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.354591316403167e-06,
+      "loss": 0.8081,
+      "step": 3358
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.329049887423198e-06,
+      "loss": 0.8028,
+      "step": 3359
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.303581926995959e-06,
+      "loss": 0.7984,
+      "step": 3360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.278187454679006e-06,
+      "loss": 0.7961,
+      "step": 3361
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.252866489973539e-06,
+      "loss": 0.8156,
+      "step": 3362
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.227619052324283e-06,
+      "loss": 0.7976,
+      "step": 3363
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.202445161119451e-06,
+      "loss": 0.2369,
+      "step": 3364
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.177344835690855e-06,
+      "loss": 0.808,
+      "step": 3365
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.152318095313778e-06,
+      "loss": 0.2315,
+      "step": 3366
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.127364959206981e-06,
+      "loss": 0.7674,
+      "step": 3367
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.102485446532689e-06,
+      "loss": 0.79,
+      "step": 3368
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.077679576396676e-06,
+      "loss": 0.7934,
+      "step": 3369
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.05294736784807e-06,
+      "loss": 0.7702,
+      "step": 3370
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.028288839879435e-06,
+      "loss": 0.7853,
+      "step": 3371
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.003704011426845e-06,
+      "loss": 0.8121,
+      "step": 3372
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.979192901369688e-06,
+      "loss": 0.7885,
+      "step": 3373
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.954755528530751e-06,
+      "loss": 0.7938,
+      "step": 3374
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.930391911676256e-06,
+      "loss": 0.7489,
+      "step": 3375
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9061020695157265e-06,
+      "loss": 0.7946,
+      "step": 3376
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.88188602070203e-06,
+      "loss": 0.8027,
+      "step": 3377
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.857743783831413e-06,
+      "loss": 0.7704,
+      "step": 3378
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8336753774434135e-06,
+      "loss": 0.8057,
+      "step": 3379
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8096808200208557e-06,
+      "loss": 0.7743,
+      "step": 3380
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.785760129989868e-06,
+      "loss": 0.7825,
+      "step": 3381
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7619133257198902e-06,
+      "loss": 0.7829,
+      "step": 3382
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.738140425523551e-06,
+      "loss": 0.782,
+      "step": 3383
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.714441447656791e-06,
+      "loss": 0.8069,
+      "step": 3384
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6908164103187316e-06,
+      "loss": 0.812,
+      "step": 3385
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.667265331651759e-06,
+      "loss": 0.7998,
+      "step": 3386
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6437882297414292e-06,
+      "loss": 0.8043,
+      "step": 3387
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6203851226165097e-06,
+      "loss": 0.7976,
+      "step": 3388
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5970560282489353e-06,
+      "loss": 0.8415,
+      "step": 3389
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5738009645538195e-06,
+      "loss": 0.7299,
+      "step": 3390
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5506199493894e-06,
+      "loss": 0.8553,
+      "step": 3391
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.527513000557059e-06,
+      "loss": 0.7878,
+      "step": 3392
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5044801358013136e-06,
+      "loss": 0.7723,
+      "step": 3393
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4815213728097928e-06,
+      "loss": 0.7931,
+      "step": 3394
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.458636729213194e-06,
+      "loss": 0.8013,
+      "step": 3395
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4358262225853254e-06,
+      "loss": 0.8652,
+      "step": 3396
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4130898704430426e-06,
+      "loss": 0.804,
+      "step": 3397
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3904276902462562e-06,
+      "loss": 0.8192,
+      "step": 3398
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3678396993979343e-06,
+      "loss": 0.8001,
+      "step": 3399
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3453259152440684e-06,
+      "loss": 0.746,
+      "step": 3400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.32288635507364e-06,
+      "loss": 0.8531,
+      "step": 3401
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3005210361186643e-06,
+      "loss": 0.7799,
+      "step": 3402
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.278229975554137e-06,
+      "loss": 0.8234,
+      "step": 3403
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2560131904980084e-06,
+      "loss": 0.7945,
+      "step": 3404
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.23387069801121e-06,
+      "loss": 0.8156,
+      "step": 3405
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2118025150976394e-06,
+      "loss": 0.8163,
+      "step": 3406
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1898086587040743e-06,
+      "loss": 0.8024,
+      "step": 3407
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1678891457202597e-06,
+      "loss": 0.8057,
+      "step": 3408
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1460439929788312e-06,
+      "loss": 0.8063,
+      "step": 3409
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.124273217255347e-06,
+      "loss": 0.8158,
+      "step": 3410
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.102576835268212e-06,
+      "loss": 0.8166,
+      "step": 3411
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.08095486367872e-06,
+      "loss": 0.7871,
+      "step": 3412
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0594073190910453e-06,
+      "loss": 0.7821,
+      "step": 3413
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.037934218052152e-06,
+      "loss": 0.771,
+      "step": 3414
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.016535577051893e-06,
+      "loss": 0.777,
+      "step": 3415
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9952114125229137e-06,
+      "loss": 0.7718,
+      "step": 3416
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9739617408406585e-06,
+      "loss": 0.7899,
+      "step": 3417
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9527865783233965e-06,
+      "loss": 0.7439,
+      "step": 3418
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.931685941232143e-06,
+      "loss": 0.7919,
+      "step": 3419
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9106598457707134e-06,
+      "loss": 0.7599,
+      "step": 3420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8897083080856703e-06,
+      "loss": 0.8412,
+      "step": 3421
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8688313442663094e-06,
+      "loss": 0.7878,
+      "step": 3422
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8480289703446737e-06,
+      "loss": 0.78,
+      "step": 3423
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8273012022955293e-06,
+      "loss": 0.7958,
+      "step": 3424
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.806648056036343e-06,
+      "loss": 0.8105,
+      "step": 3425
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7860695474272392e-06,
+      "loss": 0.7976,
+      "step": 3426
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7655656922711216e-06,
+      "loss": 0.7775,
+      "step": 3427
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7451365063134728e-06,
+      "loss": 0.8178,
+      "step": 3428
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7247820052424657e-06,
+      "loss": 0.739,
+      "step": 3429
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.704502204688941e-06,
+      "loss": 0.8022,
+      "step": 3430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6842971202263535e-06,
+      "loss": 0.8175,
+      "step": 3431
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.664166767370757e-06,
+      "loss": 0.8043,
+      "step": 3432
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6441111615808977e-06,
+      "loss": 0.7839,
+      "step": 3433
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6241303182580334e-06,
+      "loss": 0.8038,
+      "step": 3434
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6042242527460457e-06,
+      "loss": 0.8098,
+      "step": 3435
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.584392980331396e-06,
+      "loss": 0.7644,
+      "step": 3436
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.564636516243124e-06,
+      "loss": 0.8258,
+      "step": 3437
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.544954875652761e-06,
+      "loss": 0.7367,
+      "step": 3438
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5253480736744496e-06,
+      "loss": 0.8085,
+      "step": 3439
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.505816125364846e-06,
+      "loss": 0.7808,
+      "step": 3440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4863590457230743e-06,
+      "loss": 0.8187,
+      "step": 3441
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4669768496908384e-06,
+      "loss": 0.7882,
+      "step": 3442
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.447669552152265e-06,
+      "loss": 0.7528,
+      "step": 3443
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4284371679340166e-06,
+      "loss": 0.7593,
+      "step": 3444
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4092797118052236e-06,
+      "loss": 0.7572,
+      "step": 3445
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3901971984774397e-06,
+      "loss": 0.7956,
+      "step": 3446
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3711896426046987e-06,
+      "loss": 0.8059,
+      "step": 3447
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3522570587834803e-06,
+      "loss": 0.7805,
+      "step": 3448
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3333994615526433e-06,
+      "loss": 0.8049,
+      "step": 3449
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3146168653935374e-06,
+      "loss": 0.8016,
+      "step": 3450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.295909284729836e-06,
+      "loss": 0.2414,
+      "step": 3451
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2772767339276578e-06,
+      "loss": 0.7977,
+      "step": 3452
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2587192272954917e-06,
+      "loss": 0.7928,
+      "step": 3453
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.240236779084193e-06,
+      "loss": 0.7847,
+      "step": 3454
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2218294034869856e-06,
+      "loss": 0.7617,
+      "step": 3455
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.20349711463943e-06,
+      "loss": 0.8211,
+      "step": 3456
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1852399266194314e-06,
+      "loss": 0.7975,
+      "step": 3457
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.167057853447252e-06,
+      "loss": 0.8101,
+      "step": 3458
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.148950909085401e-06,
+      "loss": 0.7962,
+      "step": 3459
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1309191074387556e-06,
+      "loss": 0.8011,
+      "step": 3460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.112962462354484e-06,
+      "loss": 0.8002,
+      "step": 3461
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0950809876220113e-06,
+      "loss": 0.8278,
+      "step": 3462
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0772746969730415e-06,
+      "loss": 0.7486,
+      "step": 3463
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0595436040815817e-06,
+      "loss": 0.2385,
+      "step": 3464
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0418877225638288e-06,
+      "loss": 0.7846,
+      "step": 3465
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0243070659782705e-06,
+      "loss": 0.8132,
+      "step": 3466
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.006801647825618e-06,
+      "loss": 0.8067,
+      "step": 3467
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.989371481548807e-06,
+      "loss": 0.7665,
+      "step": 3468
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.972016580532954e-06,
+      "loss": 0.7716,
+      "step": 3469
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9547369581054074e-06,
+      "loss": 0.7808,
+      "step": 3470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9375326275357208e-06,
+      "loss": 0.8061,
+      "step": 3471
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.920403602035581e-06,
+      "loss": 0.7857,
+      "step": 3472
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.903349894758888e-06,
+      "loss": 0.8146,
+      "step": 3473
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8863715188016994e-06,
+      "loss": 0.7921,
+      "step": 3474
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8694684872021861e-06,
+      "loss": 0.7875,
+      "step": 3475
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8526408129407092e-06,
+      "loss": 0.7356,
+      "step": 3476
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8358885089397205e-06,
+      "loss": 0.7692,
+      "step": 3477
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8192115880638182e-06,
+      "loss": 0.8116,
+      "step": 3478
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.802610063119714e-06,
+      "loss": 0.8217,
+      "step": 3479
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7860839468561874e-06,
+      "loss": 0.7989,
+      "step": 3480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7696332519641312e-06,
+      "loss": 0.7963,
+      "step": 3481
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7532579910765512e-06,
+      "loss": 0.8279,
+      "step": 3482
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7369581767684774e-06,
+      "loss": 0.8423,
+      "step": 3483
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7207338215569969e-06,
+      "loss": 0.7748,
+      "step": 3484
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7045849379013102e-06,
+      "loss": 0.784,
+      "step": 3485
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6885115382026085e-06,
+      "loss": 0.7605,
+      "step": 3486
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.672513634804107e-06,
+      "loss": 0.8231,
+      "step": 3487
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.656591239991112e-06,
+      "loss": 0.8015,
+      "step": 3488
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6407443659908762e-06,
+      "loss": 0.8342,
+      "step": 3489
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6249730249726648e-06,
+      "loss": 0.8233,
+      "step": 3490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.609277229047801e-06,
+      "loss": 0.8158,
+      "step": 3491
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.593656990269521e-06,
+      "loss": 0.7702,
+      "step": 3492
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5781123206330739e-06,
+      "loss": 0.8383,
+      "step": 3493
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5626432320756667e-06,
+      "loss": 0.7979,
+      "step": 3494
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5472497364764859e-06,
+      "loss": 0.7732,
+      "step": 3495
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5319318456566423e-06,
+      "loss": 0.7945,
+      "step": 3496
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5166895713791817e-06,
+      "loss": 0.7993,
+      "step": 3497
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5015229253491302e-06,
+      "loss": 0.8206,
+      "step": 3498
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4864319192133823e-06,
+      "loss": 0.799,
+      "step": 3499
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4714165645607792e-06,
+      "loss": 0.782,
+      "step": 3500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4564768729220412e-06,
+      "loss": 0.774,
+      "step": 3501
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4416128557698028e-06,
+      "loss": 0.767,
+      "step": 3502
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4268245245185996e-06,
+      "loss": 0.8113,
+      "step": 3503
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4121118905248031e-06,
+      "loss": 0.7663,
+      "step": 3504
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3974749650867092e-06,
+      "loss": 0.7956,
+      "step": 3505
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3829137594444153e-06,
+      "loss": 0.8404,
+      "step": 3506
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3684282847799213e-06,
+      "loss": 0.782,
+      "step": 3507
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3540185522170624e-06,
+      "loss": 0.767,
+      "step": 3508
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3396845728214646e-06,
+      "loss": 0.7975,
+      "step": 3509
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3254263576006343e-06,
+      "loss": 0.8127,
+      "step": 3510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3112439175038794e-06,
+      "loss": 0.7839,
+      "step": 3511
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.29713726342231e-06,
+      "loss": 0.8333,
+      "step": 3512
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2831064061888498e-06,
+      "loss": 0.7882,
+      "step": 3513
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.269151356578202e-06,
+      "loss": 0.7858,
+      "step": 3514
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.255272125306861e-06,
+      "loss": 0.7776,
+      "step": 3515
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2414687230331123e-06,
+      "loss": 0.8172,
+      "step": 3516
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2277411603569767e-06,
+      "loss": 0.7927,
+      "step": 3517
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2140894478202657e-06,
+      "loss": 0.7824,
+      "step": 3518
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2005135959065383e-06,
+      "loss": 0.8042,
+      "step": 3519
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1870136150410883e-06,
+      "loss": 0.7796,
+      "step": 3520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1735895155909339e-06,
+      "loss": 0.7689,
+      "step": 3521
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1602413078648733e-06,
+      "loss": 0.8323,
+      "step": 3522
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1469690021133629e-06,
+      "loss": 0.8348,
+      "step": 3523
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1337726085285939e-06,
+      "loss": 0.7855,
+      "step": 3524
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1206521372445044e-06,
+      "loss": 0.7774,
+      "step": 3525
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1076075983366573e-06,
+      "loss": 0.7958,
+      "step": 3526
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0946390018223506e-06,
+      "loss": 0.8166,
+      "step": 3527
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0817463576605513e-06,
+      "loss": 0.7862,
+      "step": 3528
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0689296757519063e-06,
+      "loss": 0.7819,
+      "step": 3529
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0561889659387203e-06,
+      "loss": 0.7673,
+      "step": 3530
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0435242380049559e-06,
+      "loss": 0.759,
+      "step": 3531
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0309355016762557e-06,
+      "loss": 0.8125,
+      "step": 3532
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0184227666198643e-06,
+      "loss": 0.8165,
+      "step": 3533
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0059860424446842e-06,
+      "loss": 0.8085,
+      "step": 3534
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.936253387012318e-07,
+      "loss": 0.7769,
+      "step": 3535
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.813406648816804e-07,
+      "loss": 0.7905,
+      "step": 3536
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.691320304197837e-07,
+      "loss": 0.8152,
+      "step": 3537
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.569994446909091e-07,
+      "loss": 0.8069,
+      "step": 3538
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.44942917012026e-07,
+      "loss": 0.7993,
+      "step": 3539
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.329624566417172e-07,
+      "loss": 0.7997,
+      "step": 3540
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.21058072780101e-07,
+      "loss": 0.7792,
+      "step": 3541
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.092297745689426e-07,
+      "loss": 0.8105,
+      "step": 3542
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.974775710915207e-07,
+      "loss": 0.7877,
+      "step": 3543
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.858014713726936e-07,
+      "loss": 0.7961,
+      "step": 3544
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.742014843789004e-07,
+      "loss": 0.7969,
+      "step": 3545
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.626776190181041e-07,
+      "loss": 0.817,
+      "step": 3546
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.512298841398259e-07,
+      "loss": 0.7256,
+      "step": 3547
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.398582885351225e-07,
+      "loss": 0.7818,
+      "step": 3548
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.285628409365753e-07,
+      "loss": 0.7747,
+      "step": 3549
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.173435500182902e-07,
+      "loss": 0.8323,
+      "step": 3550
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.062004243958865e-07,
+      "loss": 0.7758,
+      "step": 3551
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.951334726265191e-07,
+      "loss": 0.8159,
+      "step": 3552
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.841427032088122e-07,
+      "loss": 0.7876,
+      "step": 3553
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.732281245829142e-07,
+      "loss": 0.2441,
+      "step": 3554
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.623897451304318e-07,
+      "loss": 0.7901,
+      "step": 3555
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.51627573174507e-07,
+      "loss": 0.7867,
+      "step": 3556
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.409416169797067e-07,
+      "loss": 0.2322,
+      "step": 3557
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.303318847521001e-07,
+      "loss": 0.79,
+      "step": 3558
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.197983846392253e-07,
+      "loss": 0.8104,
+      "step": 3559
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.093411247300452e-07,
+      "loss": 0.7922,
+      "step": 3560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.98960113055025e-07,
+      "loss": 0.8224,
+      "step": 3561
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.886553575860322e-07,
+      "loss": 0.7636,
+      "step": 3562
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.784268662363924e-07,
+      "loss": 0.7785,
+      "step": 3563
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.682746468608781e-07,
+      "loss": 0.758,
+      "step": 3564
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.58198707255675e-07,
+      "loss": 0.8099,
+      "step": 3565
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.481990551583938e-07,
+      "loss": 0.8383,
+      "step": 3566
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.382756982480587e-07,
+      "loss": 0.2613,
+      "step": 3567
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.284286441451293e-07,
+      "loss": 0.7879,
+      "step": 3568
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.186579004114123e-07,
+      "loss": 0.7847,
+      "step": 3569
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.089634745501838e-07,
+      "loss": 0.8407,
+      "step": 3570
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.993453740060772e-07,
+      "loss": 0.7382,
+      "step": 3571
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.898036061650957e-07,
+      "loss": 0.7818,
+      "step": 3572
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.803381783546669e-07,
+      "loss": 0.8141,
+      "step": 3573
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.709490978435539e-07,
+      "loss": 0.8234,
+      "step": 3574
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.616363718419226e-07,
+      "loss": 0.7736,
+      "step": 3575
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.52400007501297e-07,
+      "loss": 0.7674,
+      "step": 3576
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.432400119145364e-07,
+      "loss": 0.8247,
+      "step": 3577
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.34156392115881e-07,
+      "loss": 0.7912,
+      "step": 3578
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.251491550809062e-07,
+      "loss": 0.8111,
+      "step": 3579
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.16218307726557e-07,
+      "loss": 0.8081,
+      "step": 3580
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.073638569110695e-07,
+      "loss": 0.8102,
+      "step": 3581
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.985858094340379e-07,
+      "loss": 0.2454,
+      "step": 3582
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.898841720364144e-07,
+      "loss": 0.8065,
+      "step": 3583
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.812589514004095e-07,
+      "loss": 0.7718,
+      "step": 3584
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.727101541496026e-07,
+      "loss": 0.7977,
+      "step": 3585
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6423778684886455e-07,
+      "loss": 0.8046,
+      "step": 3586
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.558418560043909e-07,
+      "loss": 0.7847,
+      "step": 3587
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4752236806363535e-07,
+      "loss": 0.7706,
+      "step": 3588
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.392793294153985e-07,
+      "loss": 0.7927,
+      "step": 3589
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.311127463897502e-07,
+      "loss": 0.8167,
+      "step": 3590
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.230226252580516e-07,
+      "loss": 0.7962,
+      "step": 3591
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.150089722329442e-07,
+      "loss": 0.8059,
+      "step": 3592
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.07071793468361e-07,
+      "loss": 0.7548,
+      "step": 3593
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9921109505947075e-07,
+      "loss": 0.7853,
+      "step": 3594
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.91426883042767e-07,
+      "loss": 0.7732,
+      "step": 3595
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8371916339594585e-07,
+      "loss": 0.7765,
+      "step": 3596
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.76087942038017e-07,
+      "loss": 0.7801,
+      "step": 3597
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6853322482921505e-07,
+      "loss": 0.7527,
+      "step": 3598
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6105501757102145e-07,
+      "loss": 0.7861,
+      "step": 3599
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5365332600617586e-07,
+      "loss": 0.7843,
+      "step": 3600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4632815581866484e-07,
+      "loss": 0.8123,
+      "step": 3601
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3907951263371095e-07,
+      "loss": 0.7692,
+      "step": 3602
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.319074020177282e-07,
+      "loss": 0.7909,
+      "step": 3603
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.248118294784219e-07,
+      "loss": 0.7913,
+      "step": 3604
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.177928004646891e-07,
+      "loss": 0.7929,
+      "step": 3605
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.108503203666402e-07,
+      "loss": 0.7764,
+      "step": 3606
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0398439451561066e-07,
+      "loss": 0.835,
+      "step": 3607
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.971950281841607e-07,
+      "loss": 0.8132,
+      "step": 3608
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.904822265860307e-07,
+      "loss": 0.7825,
+      "step": 3609
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8384599487618625e-07,
+      "loss": 0.8079,
+      "step": 3610
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7728633815079507e-07,
+      "loss": 0.8412,
+      "step": 3611
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7080326144719447e-07,
+      "loss": 0.7716,
+      "step": 3612
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6439676974393537e-07,
+      "loss": 0.7665,
+      "step": 3613
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.580668679607601e-07,
+      "loss": 0.7797,
+      "step": 3614
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5181356095858033e-07,
+      "loss": 0.2222,
+      "step": 3615
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4563685353949925e-07,
+      "loss": 0.8121,
+      "step": 3616
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3953675044680046e-07,
+      "loss": 0.8013,
+      "step": 3617
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.335132563649145e-07,
+      "loss": 0.7975,
+      "step": 3618
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.275663759194857e-07,
+      "loss": 0.7901,
+      "step": 3619
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2169611367727215e-07,
+      "loss": 0.7982,
+      "step": 3620
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1590247414624566e-07,
+      "loss": 0.7468,
+      "step": 3621
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.101854617754917e-07,
+      "loss": 0.8372,
+      "step": 3622
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0454508095528734e-07,
+      "loss": 0.7797,
+      "step": 3623
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.989813360170456e-07,
+      "loss": 0.7824,
+      "step": 3624
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9349423123332654e-07,
+      "loss": 0.8141,
+      "step": 3625
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8808377081785956e-07,
+      "loss": 0.7562,
+      "step": 3626
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8274995892546555e-07,
+      "loss": 0.8376,
+      "step": 3627
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7749279965215692e-07,
+      "loss": 0.7644,
+      "step": 3628
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7231229703505992e-07,
+      "loss": 0.7659,
+      "step": 3629
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6720845505242555e-07,
+      "loss": 0.779,
+      "step": 3630
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6218127762365198e-07,
+      "loss": 0.8077,
+      "step": 3631
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5723076860927333e-07,
+      "loss": 0.8006,
+      "step": 3632
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5235693181090417e-07,
+      "loss": 0.7991,
+      "step": 3633
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4755977097131724e-07,
+      "loss": 0.788,
+      "step": 3634
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4283928977441018e-07,
+      "loss": 0.8136,
+      "step": 3635
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3819549184516112e-07,
+      "loss": 0.7868,
+      "step": 3636
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3362838074967298e-07,
+      "loss": 0.808,
+      "step": 3637
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2913795999519586e-07,
+      "loss": 0.7869,
+      "step": 3638
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.24724233030038e-07,
+      "loss": 0.8181,
+      "step": 3639
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2038720324362152e-07,
+      "loss": 0.7663,
+      "step": 3640
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1612687396650445e-07,
+      "loss": 0.7584,
+      "step": 3641
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1194324847030313e-07,
+      "loss": 0.8179,
+      "step": 3642
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0783632996774761e-07,
+      "loss": 0.8508,
+      "step": 3643
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0380612161265956e-07,
+      "loss": 0.8098,
+      "step": 3644
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.98526264999633e-08,
+      "loss": 0.2521,
+      "step": 3645
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.597584766566359e-08,
+      "loss": 0.824,
+      "step": 3646
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.217578808684568e-08,
+      "loss": 0.8243,
+      "step": 3647
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.84524506817086e-08,
+      "loss": 0.7896,
+      "step": 3648
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.480583830948741e-08,
+      "loss": 0.8046,
+      "step": 3649
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.123595377053095e-08,
+      "loss": 0.2425,
+      "step": 3650
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.774279980626853e-08,
+      "loss": 0.7833,
+      "step": 3651
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.432637909918772e-08,
+      "loss": 0.7938,
+      "step": 3652
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.098669427286764e-08,
+      "loss": 0.7824,
+      "step": 3653
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.772374789194569e-08,
+      "loss": 0.2143,
+      "step": 3654
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.453754246212861e-08,
+      "loss": 0.7904,
+      "step": 3655
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.142808043020365e-08,
+      "loss": 0.7356,
+      "step": 3656
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.839536418401625e-08,
+      "loss": 0.8082,
+      "step": 3657
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.543939605249238e-08,
+      "loss": 0.8481,
+      "step": 3658
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.256017830559401e-08,
+      "loss": 0.7704,
+      "step": 3659
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.975771315435252e-08,
+      "loss": 0.7931,
+      "step": 3660
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7032002750879714e-08,
+      "loss": 0.7675,
+      "step": 3661
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.438304918831238e-08,
+      "loss": 0.8111,
+      "step": 3662
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.181085450087885e-08,
+      "loss": 0.7332,
+      "step": 3663
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9315420663832426e-08,
+      "loss": 0.764,
+      "step": 3664
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.689674959348466e-08,
+      "loss": 0.8072,
+      "step": 3665
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4554843147216464e-08,
+      "loss": 0.8198,
+      "step": 3666
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.228970312343371e-08,
+      "loss": 0.7719,
+      "step": 3667
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.010133126162273e-08,
+      "loss": 0.8079,
+      "step": 3668
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7989729242294814e-08,
+      "loss": 0.7977,
+      "step": 3669
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.59548986869973e-08,
+      "loss": 0.8058,
+      "step": 3670
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3996841158346882e-08,
+      "loss": 0.7928,
+      "step": 3671
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2115558159985226e-08,
+      "loss": 0.8044,
+      "step": 3672
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.031105113662335e-08,
+      "loss": 0.773,
+      "step": 3673
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8583321473986116e-08,
+      "loss": 0.7873,
+      "step": 3674
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.693237049883445e-08,
+      "loss": 0.7898,
+      "step": 3675
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.535819947900974e-08,
+      "loss": 0.7787,
+      "step": 3676
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3860809623345017e-08,
+      "loss": 0.7663,
+      "step": 3677
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2440202081731578e-08,
+      "loss": 0.7722,
+      "step": 3678
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.109637794509677e-08,
+      "loss": 0.772,
+      "step": 3679
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.829338245403997e-09,
+      "loss": 0.8031,
+      "step": 3680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.639083955663818e-09,
+      "loss": 0.7623,
+      "step": 3681
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.525615989889546e-09,
+      "loss": 0.8058,
+      "step": 3682
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.488935203152746e-09,
+      "loss": 0.2405,
+      "step": 3683
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.529042391549944e-09,
+      "loss": 0.7703,
+      "step": 3684
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.645938292213714e-09,
+      "loss": 0.7739,
+      "step": 3685
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.839623583301588e-09,
+      "loss": 0.2393,
+      "step": 3686
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.110098884007151e-09,
+      "loss": 0.2755,
+      "step": 3687
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.457364754560043e-09,
+      "loss": 0.7964,
+      "step": 3688
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.881421696214858e-09,
+      "loss": 0.7686,
+      "step": 3689
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3822701512400394e-09,
+      "loss": 0.8049,
+      "step": 3690
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.599105029622911e-10,
+      "loss": 0.7884,
+      "step": 3691
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.14343075722168e-10,
+      "loss": 0.7616,
+      "step": 3692
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4556813489627914e-10,
+      "loss": 0.8025,
+      "step": 3693
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5358588687508502e-10,
+      "loss": 0.8219,
+      "step": 3694
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.839647908510102e-11,
+      "loss": 0.8089,
+      "step": 3695
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.8339,
+      "step": 3696
+    },
+    {
+      "epoch": 1.0,
+      "step": 3696,
+      "total_flos": 1.27730137104384e+16,
+      "train_loss": 0.7962456708854605,
+      "train_runtime": 177781.5934,
+      "train_samples_per_second": 3.742,
+      "train_steps_per_second": 0.021
+    }
+  ],
+  "max_steps": 3696,
+  "num_train_epochs": 1,
+  "total_flos": 1.27730137104384e+16,
+  "trial_name": null,
+  "trial_params": null
+}